For The Warehouse Group a missing shoe buyer cost the company millions, with declining sales, weaker profitability, stock markdowns and lost market share as product assortments went largely unchanged for two years.

“Where you have the real risk in the organisation is your people and processes – and that’s where the unstructured data lives.”

The experience convinced group chief executive Mark Stirton that one of the biggest risks facing organisations isn’t a lack of data, but a lack of accessible knowledge, and it’s helped shape the Kiwi retailer’s thinking about data, knowledge and emerging AI technologies.

Speaking at the recent Snowflake World Tour Auckland, Stirton outlined how the company is exploring what he calls a ‘digital bench’ and an ‘agentic university’ as part of a broader effort to make organisational knowledge easier to access and less dependent on individual employees.

Stirton was blunt that the retail game is changing. “If you don’t think its changing, I think you might be under a rock. And when the game’s changing, you’ve got to decide what are the new rules? And what is the new playbook you need to run these rules against?”

He described what he sees as the next major shift in business. The first wave gave organisations access to manufacturing, infrastructure and eCommerce capabilities on demand. The next, he says is ‘pay-as-you-go intelligence’.

“When you think about tokens and all these models, you’re getting intelligence at scale, which has never happened in the history of man.”

The belief is influencing how the company is preparing for the future.

Stirton says he has challenged leaders across the business to rethink how their departments operate, saying ’30 percent of it must change’ as the organisation prepares for changing business and operating models.

The catalyst isn’t a new technology – not even AI. Instead it’s that recognition that critical expertise often sits inside people’s heads.

In the footwear example, the absence of a buyer did not simply create a vacancy, it triggered a chain reaction as assortments were not refreshed and decisions were delayed and effects flowed on to other areas of the team and business.

“It’s the contagion of small risks. They don’t look big in isolation but when you add them up they start to look like a really destructive force.”

Building a digital bench…

The challenge sits behind one of the more unusual concepts discussed during the presentation – that of a ‘digital bench’.

Using a rugby analogy, Grayston suggested many companies are trying to deliver the same outcomes despite effectively playing short-handed. Businesses expect results even when expertise is missing, teams are understaffed or key knowledge exists only with a handful of employees.

The digital bench is his attempt to address that problem.

Rather than relying on individual employees to hold critical information, Stirton wants organisational knowledge to become more accessible across the enterprise. The goal is not to simply store information, but to make expertise available when people need it.

That ambition is shaping the retailer’s technology foundations.

The Warehouse is working with Snowflake on semantic layers and vector-based capabilities, while also experimenting with Cortex to surface insights from existing dashboards and data assets. Stirton describes this work as the ‘unglamorous’ foundations required before organisations can take advantage of future capabilities.

He argues that structured enterprise data is only part of the picture.

While information stored in systems such as ERP platforms can be managed and analysed, many organisational risks sit elsewhere.

“But where you have the real risk in the organisation is your people and processes – and that’s where the unstructured data lives and where most of the gaps start to form when you lose a person.”

…And an agentic university

That thinking has given rise to an even bigger concept: what Stirton calls an ‘agentic university’.

Universities, he noted, create knowledge, preserve knowledge and transmit knowledge. In his view, businesses increasingly need to do the same thing.

He described a future where employees can work alongside knowledge layers and context layers that help connect information from across the organisation. Rather than expertise being concentrated within individuals, it would become easier to access and share.

“There would be abundance of knowledge and those knowledge gaps where there’s no training or there’s no parts where knowledge is transferred, that starts to be on tap for someone and then connected brains start to connect the whole organisation.

“So if I’m a buyer and I want to ask a question on my supply chain, suddenly you’ve actually got a knowledge layer – ‘your professors with your long-term knowledge’ and a context layer that now connects with each other.”

Stirton, who admits he’s ‘a bit of a dreamer and what I call a metaphor person’ referred to the concept as a corporate ‘second brain’. The idea is that organisational memory, historical decisions, operational context and expertise become easier to draw upon regardless of where they originate.

Stirton is not approaching the challenge as a company-wide technology rollout, saying he’s avoiding the ‘spray and pray’ approach that leads to burning tokens.

He says the retailer is focusing on innovators and early adopters – the 2.5 percent who ‘statistically will always move the organisation faster because they’re the ones who are braver, more courageous and curious’  arguing that change rarely happens evenly across an organisation. These groups, he believes, can help accelerate adoption and encourage new ways of working across the wider business.

“Adoption will always be uneven in technology,” he says.

“I think the people with the highest desire to have a growth mindset will start to use the technology and the knowledge layer and the context layer at much more rapid speed.”

The result is a view of AI that differs from many enterprise discussions focused on productivity gains or automation.

Instead, Grayston’s presentation centred on something more fundamental: knowledge.

Who owns it. Where it lives. What happens when it disappears.

The footwear example offered a costly reminder of the consequences when critical expertise resides with one person. The digital bench and agentic university are attempts to ensure that knowledge becomes an organisational asset rather than an individual one.

The post The Warehouse’s empty chair wake-up call appeared first on iStart keeping business informed on technology.

The comment from Nastassia Subritzky elicited plenty of laughs at Snowflake’s World Tour Auckland, but the Mitre 10 head of data and insights admits she was ‘pretty alarmed’ when she did the count during the hardware chain’s migration to Fabric.

“Our human users are also getting trained by our agents.”

“This reporting sprawl, plus the seemingly infinite number of spreadsheets makes it difficult because you’ve got all these different reports and spreadsheets and they all tend to have slightly different numbers, so you end up spending so much time trying to reconcile between all of them,” Subritzky says.

The problem arrived at the worst possible time, with the co-op already in the middle of a major ERP transformation, moving from individually customised store data warehouses to a centrally managed SAP platform and data warehouse.

“The move to a centrally managed data warehouse has naturally resulted in many cries for help with how to actually analyse data in this new SAP world,” she says.

Stores were struggling to work out where to go find the analysis they previously had available. The challenge was compounded by Mitre 10’s cooperative structure, where independently owned stores have their own approaches to performance and operations.

And then there was Excel.

“Turns out people get really attached to their Excel spreadsheets, especially during times of change,” she says.

Breaking free from report sprawl

Rather than create report number 3,065, the retailer decided to do something different.

The goal wasn’t to replace reporting, but to reduce dependence on it through AI agents built on curated datasets. Using Snowflake CoWork, the company created agents that draw from the same underlying data as Power BI, ensuring answers remain tied to a common source of truth.

The governance layer was critical.

Many organisations are discovering that generative AI amplifies existing data problems. Mitre 10 started with carefully controlled datasets, guardrails and verified queries to prevent users from generating yet another collection of disconnected answers. The retailer also maintained control over which datasets agents could access.

“With Snowflake Co-work, we have complete control over what datasets the agents have access to, and naturally to help with some reconciliation issues they use the same base tables as our Power BI workbooks.

“Everything naturally ties back together which has been a fantastic win for manual work reduction because I don’t have to send my engineers out to reconcile yet another Excel spreadsheet.”

Trent: Good tech skills, no business sense

The first AI agent, Trent, is the merchandising team’s ‘general assistant’. He was rolled out to the team as part of a two month deadline-driven migration from a legacy SQL reporting platform.

“This deadline doesn’t sound too bad until you consider how many workbooks someone can connect to a very old SQL reporting server over the course of a decade,” Subritzky quips. “We thought given they would have to get used to a new way of working anyway, why not just chuck them in the deep end and give them something really, really different to worry about?”

Trent helps with common tasks, including looking at supplier performance, undertaking basic rent reviews and looking at product performance by store.

But launching the agent was just the start.

“We got basic Trent up and running in absolutely no time at all, but he was very much like an intern. Really good technical skills, but absolutely no business knowledge whatsoever.”

What followed looks remarkably similar to onboarding a new analyst.

The team reviewed logs, refined prompts, built verified queries and taught the agent how the business actually worked. Instructions were added to standardise responses, eliminate exaggeration and ensure consistent analytical outputs.

Trent’s creator and Subritzky reviewed logs created from the merchandising team’s enthusiastic testing of Trent, turning successful requests into verified queries. There was some fun along the way too, with easter eggs added for testers.

The effort paid off.

“Trent had gone from being a really enthusiastic intern to a competent junior analyst who sometimes surprises us with some quite cool solutions.”

Today the agent is embedded in the merch teams daily workflow and has helped staff focus their analysis rather than build ever-larger spreadsheets.

Agents training humans

The store pilot provided more insights.

Initially, staff asked relatively simple questions. Over time, as confidence grew and the results were verified, the complexity of those questions increased. Store operators moved from asking basic sales questions to exploring inventory optimisation, promotional effectiveness and slow-moving stock analysis.

One example involved a store asking which products should be included in a winter clearance programme for slow and obsolete stock. Instead of simply producing raw figures, the agent asked clarifying questions, identified priority categories and recommended potential actions.

“What we were finding here was they could actually genuinely chat away to it because my team had done all the hard work up front to get all of those table joins going, all of the Mitre 10 context, all the odd little language that you end up having in business.”

Just as interesting was what happened to the users.

Agents were instructed to push back on vague or incomplete requests and ask for more information when necessary.

“In the interest of the cost, the compute time, and user experience, we added in explicit instructions about what sort of prompts the agent should go back to the user with if they didn’t get enough information.”

If users keep asking ambiguous questions, the agent keeps prompting them for more information.

Over time, users learned to write better questions.

“So our human users are also getting trained by our agents.”

That improvement in data literacy surfaced elsewhere as well. Because the platform could return the SQL underpinning its answers, analysts began developing a stronger understanding of how data tables connected and how metrics were calculated. The result was not simply more automation, but more capable users.

Meanwhile, executives discovered a different benefit.

With access to mobile capabilities, senior leaders began using the agents directly during meetings. Questions that previously required analyst support could be answered immediately. In stores, managers started interrogating data from the shop floor, investigating anomalies as they encountered them rather than waiting to return to a laptop.

Most importantly, the initiative appears to be delivering against its original objective. During the pilots, stores participating in the programme requested fewer new Power BI reports and modifications. Users were finding answers themselves rather than generating more reporting demand.

As Mitre 10 looks ahead, the challenge is ensuring history doesn’t repeat itself.

The retailer is already receiving requests for additional AI agents and is wary of replacing one form of sprawl with another. “We’re working out how we can make the existing analysts flexible, what sort of combinations we can use so that we can reuse a lot of what we’ve got and only do new agents when it really makes sense or is a really separate use case.”

After all, nobody wants to spend next year cleaning up 3,064 agents.

The post Death by Power BI? Mitre 10 fights back appeared first on iStart keeping business informed on technology.

The move was quickly followed by New Zealand’s equivalent, Inland  Revenue, releasing draft guidance on how software payments, cloud services and SaaS subscriptions should be treated for tax purposes.

“The ATO takes a broader interpretation of when a royalty may arise in cloud-computing context.”

But while both countries are trying to update rules written long before the cloud era, they appear to be taking different approaches.

Under the ATO’s new ruling, TR 2026/2, some software and SaaS payments may be classified as royalties rather than ordinary commercial transactions, exposing a wider range of cross-border software arrangements to royalty withholding tax of between five percent and 15 percent. It’s a ruling aimed primarily at multinational digital giants such as Amazon, Microsoft, Meta, Google and Netflix, and distributors who structure intellectual property and revenue flows across multiple jurisdictions, and will see income earned from cloud and streaming services become taxable in Australia for the first time beyond GST.

By contrast, the Inland Revenue draft guidance, issued earlier this month and updating a framework dating back to 2003, largely maintains the view that most straightforward cloud software subscriptions do not give rise to royalty payments because customers are buying access to software, not the right to exploit the underlying IP. Consultation is currently underway on the draft interpretation guideline PUB00266.

“The ATO takes a broader interpretation of when a royalty may arise in cloud-computing context,” Deloitte New Zealand noted in a recent tax alert.

In both countries, global tech giants generate substantial revenue – billions in Australia’s case particularly – from local customers, yet the profits ultimately taxed in the country have often been significantly more modest after accounting for distribution and service fees and other cross-border payments within their corporate structures.

Chasing profits, not GST

For years, governments have focused on collecting GST from foreign suppliers of digital services. Australia and New Zealand both require many offshore providers to collect GST on products ranging from streaming subscriptions and digital advertising to cloud services and software licences.

GST, however, was only ever part of the solution.

When businesses buy cloud hosting, software subscriptions or online advertising, the GST paid is generally reclaimed through the normal input tax process. Governments collect the tax, but much of it is ultimately returned. The larger question has been how to tax the profits generated from those transactions when the underlying intellectual property is owned offshore.

That’s a debate that has simmered for years as the global tech giants generated billions in revenue while reporting far smaller local – taxable – profits.

The ATO’s new ruling is an attempt to address part of that issue as it targets billions in untaxed offshore software royalties.

A broader interpretation

The new ruling broadens the circumstances in which software-related payments can be treated as royalties for tax purposes. Jethro Byrne, Grant Thornton corporate tax partner, says the ATO has formed the view that Australian distributors ‘effectively posses the rights to on-sell or distribute software, which in its view is generally a royalty’.

 According to the ATO, payments may qualify as royalties when they are made for the use of, or right to use, copyright or similar intellectual property rights. The ruling points to situations where software intermediaries reproduce, communicate, modify or adapt software, or otherwise exercise rights associated with copyright ownership. It also addresses modern software distribution and SaaS arrangements.

The ATO has also made clear that it is particularly concerned about cross-border arrangements that may reduce or avoid tax on profits connected with Australia – an issue that has also long vexed many countries, including New Zealand.

For multinational software vendors, distributors and cloud providers, the Australian moves raise the prospect of much closer examination of how software revenues are structured and reported. Professional advisers have already warned them to reassess software licensing arrangements, treaty positions, withholding tax obligations and transfer pricing documentation in light of the ruling.

US pushback 

Australia’s tougher stance hasn’t been universally welcomed and has put it at odds with some of its biggest trading partners.

Previous drafts of the ruling drew criticism from the US Treasury and the US National Foreign Trade Council business association argues the ruling ignores recent Australian case law and ‘further erodes investor confidence as well as the overall business climate in the country’.

One of the more contentious aspects of the ATO’s position is that the ruling applies to payments made both before and after its publications. The ATO maintains the ruling reflects its longstanding interpretation of the law rather than a policy change, a position that could expose some existing arrangements to review.

Critics also argue that Australia’s interpretation of software royalties is pushing beyond internationally accepted practice.

Extensions ahead?

The Australian Institute of Company Directors meanwhile, is warning that while the ATO’s ruling targets software and tech, ‘it’s implications could soon reach almost all Australian industries’.

“While end-user companies aren’t directly impacted, boards should be aware the new tax will likely be built into the future pricing of their software subscriptions,” it says, while also warning that other industries with value tied to intangible assets could be future targets. It suggests sectors including pharmaceuticals, cosmetics, motor vehicles and beyond could see similar attention in future.

The post Australia declares war on software tax loopholes as NZ takes it softly appeared first on iStart keeping business informed on technology.

While consumer AI tools compete to be more conversational and creative, businesses are clear that accuracy, governance and trust often matter more then delivering a flashy user experience.

Speaking at Gartner’s IT Symposium/Xpo on the Gold Coast recently, Rutter – who is executive manager, programs and systems within Enterprise Risk at Suncorp – and AGL head of architecture – security and corporate technology Yaso Addanki, shared how they are embedding AI into their businesses, moving away from AI experimentation towards practical business applications designed to solve specific operational problems.

Stopping the bottleneck

At AGL, AI agents were built not as part of a grand transformation strategy, but because security reviews were slowing projects down. As AI initiatives multiplied across the business, security architects found themselves reviewing a growing number of projects, each requiring assessments, threat modelling and architecture reviews before work could proceed.

“We were thinking how do we not become a bottleneck and actually help the organisation move faster with deploying their initiatives?,” Addanki says.

“We had internal discussions and thought ‘how much time are we taking to turn around a security architecture review and how can we make it better and faster?’”

The discussions evolved into the development of an AI agent which uses threat modelling frameworks to generate security review documentation, providing architects with an initial draft they can refine, rather than creating assessments from scratch.

Initially deployed within the security architecture team, the agent was designed to reduce turnaround times and free specialists from routine work. But AGL sees a bigger opportunity ahead.

Rather than waiting for projects to reach security teams, the company is exploring how the technology could be made available directly to solution architects, allowing controls to be incorporated much earlier in the design process. The goal is to improve the organisation’s overall security posture while helping delivery teams move faster.

The project reflects a broader approach to AI across AGL. Addanki says the company is focusing on three key areas: Building enterprise AI and data platforms, leveraging AI capabilities already embedded in existing tech tools and orchestrating those capabilities into agents that support business workflows.

Customer service has been one of the beneficiaries. Within one AGL subsidiary, AI is used to generate summaries of customer interactions, recommend relevant knowledge articles, draft email responses and deliver more consistent service across teams. By removing the need for staff to manually document every interaction, customer service representatives can spend more time focusing on customers rather than admin.

Teaching AI to speak governance

Suncorp has taken a similar approach, focusing on solving specific business problems within risk management.

“We started with business problems first,” Rutter says. “Things that we would have looked to provide an enhancement solution for pre-AI. So these things were areas that we were going to address regardless.”

“We’ve got three areas we like to tackle: UX, efficiency and then the data quality,” Rutter says.

One project focused on incident reporting. Whenever an operational risk incident occurs, information needs to be captured accurately so teams can investigate causes, resolve issues and prevent similar incidents in future. While reports are usually submitted by risk specialists they can also be submitted by frontline employees who don’t have experience in lodging incident reports.

Suncorp’s AI-assisted incident process guides users through submissions, prompting them for missing information and helping categorise incidents correctly.

“If you haven’t provided enough, it’ll prompt you for what you’ve missed and it’ll categorise it and help it flow through our process effectively.”

The result is improved data quality from the outset, Rutter says, making downstream investigations and reporting more effective. “It’s [an area] where we will get efficiencies, but data quality is the driver.”

The company is also using AI to tackle another common enterprise problem: Dense policy documentation.

Risk and governance documents are often lengthy, technical and difficult for employees to navigate. Suncorp is using AI to simplify the language, reduce duplication and then serve that information up in a chatbot-style interface that enables staff to ask questions directly, rather than searching through extensive document libraries.

A third initiative applies AI to control testing and governance activities, helping automate administrative tasks that would otherwise consume significant amounts of specialist time. Again, the objective is not workforce reduction, but enabling experts to focus on work that requires human judgement.

Why enterprise AI can’t act like ChatGPT

Despite the growing use of AI, both organisations say the path hasn’t been without challenges – including the issue of trust. And here’s where Rutter’s comment’s about being a ‘less than ChatGPT experience’ come in.

“In the example of the policy consumption, one of the biggest risks we’ve got is the accuracy of information. It somebody has asked our assistant for support and it doesn’t give them the right answer back, we’re in a bad position.”

The company has built guardrails into its systems, including source references, hyperlinks back to supporting material and explanations showing how answers were generated. Inbuilt guardrails behind the scenes ensure the chatbot won’t hallucinate or go off topic.

The approach may create a more constrained experience than employees get from tools such as ChatGPT, but it ensures there’s no providing of wrong information.

“Something we’ve learned along the way is people will expect that [ChatGPT experience] and we’ve had to educate them that this is serving a specific purpose and there are controls in place.”

At AGL, meanwhile, security teams face a different challenge: Keeping pace with rapidly evolving models and vendor offerings. Understanding where models are deployed, where data is stored, whether information is used for training and what controls are needed around those environments has become an increasingly important part of AI governance, Addanki says.

“Sometimes it could become a black box because we don’t understand and some vendors might not be able to kind of reveal some of those details to us,” she says.

She notes that could see contractual controls becoming required.

Winning buy-in

Both organisations also highlighted the importance of proving value early. At Suncorp, proof-of-concept projects have played a key role in building stakeholder support and demonstrating how AI can address real business pain points. AGL similarly started by solving problems within its own security function before expanding further.

For both, success has come from applying AI to specific operational challenges, removing administrative burden and helping specialists focus on the work that matters most. As Richie Paul, IBM generative AI practice lead, who hosted the session, summed up: “Yasso and Jonathan give us insight into this idea of having multiple programs going in parallel and really, a thousand flowers truly will bloom, which will be the marker of enterprise-wide organisational change.”

The post AGL and Suncorp on building AI that’s worse than ChatGPT appeared first on iStart keeping business informed on technology.

For English, CEO of Velocity Global, it’s not the AI moonshots dominating conference agendas which will bring real value to business, but instead the technology’s impact on the hundreds of small administrative tasks that quietly consume time across purchasing, inventory management, customer service and approvals. As businesses push for more AI capabilities from their ERP platforms, English says many are discovering the biggest opportunities are often hidden in the operational friction they’ve learned to live with.

English says demand for AI is no longer the issue.

“There’s often an impasse where a customer says, ‘just make us AI’, which is a little too broad and untethered to their world for it ever to be successful.”

“Almost all the customers we’ve sat down with are definitely seeing opportunity in AI,” he told iStart. But he notes “I don’t think that’s necessarily crystallised to specific use cases.”

Velocity Global, which implements and supports Pronto Software, works predominantly with organisations in manufacturing, logistics and distribution, typically employing between 50 and 250 staff. Across those sectors, English says interest in AI is widespread, particularly around productivity, forecasting, inventory management and customer service.

The challenge is turning that interest into something practical. Alongside issues with quality of data – “The system is only ever going to be as good as what you put into it” – he sees being clear on use cases as the biggest issue facing companies.

“There’s often an impasse where a customer says, ‘just make us AI’, which is kind of a little bit too broad and untethered to their world for it ever to be successful.”

Instead, he argues organisations need to approach AI much like any other technology investment by identifying where people spend time, where processes stall and where repetitive tasks create friction.

“Like any automation project, where is the time going? What’s the best bang for buck? What does the technology most align to in terms of driving efficiencies or productivity gains? And how do you leverage those gains to reinforce and build on the wins?

“You start with your low-hanging fruit.”

Early wins

English’s comments come on the back of Pronto’s latest AI roadmap announcement, which coincides with the 50th year of Pronto ERP in the A/NZ market. The announcement includes new and enhanced AI capabilities including natural language interaction, predictive inventory demand and fulfilment forecasting and voice enabled BI reporting and dashboards, across Pronto Xi. Other capabilities include machine learning predictions for service and maintenance, agentic assisted purchase order approval, automated inventory checks and stock recommendations, and intelligent B2B eCommerce agents.

English says addition of a more conversational element to getting data out of the ERP system is likely to be the biggest winner for customers near term.

“If you wanted to ask how much of a given stock item is in stock, you’d have to navigate through layers of menus. Now, there’ll be an app and you can say ‘tell me how much stock of X do I have? And it will do a live inquiry on your data and come back with an answer.”

That capability is as much about reducing friction as it is about introducing new technology.

“For your typical operational managers, it just makes finding answers quicker, which is a reduction in friction,” English says.

“The other big boon will be at the executive layer, where typically your CFOs, your CEOs, your COOs, they’re not in the system day in and day out.

“They’ll be asking a subordinate for that information. It provides a new interface to the system to surface up those answers in real time.

Another early use case courtesy of Pronto is purchase order approvals.

Today, approvals often require users to log into a system, navigate to the relevant screen and manually complete the transaction. In future, many of those interactions could be reduced to a simple response delivered through an AI-driven workflow.

“You’ll get a message saying, ‘Hey, you’ve got this PO’, and you can just respond yes, and then that goes back to actually do the mechanism of approving the order in the system without you needing to go through that step of logging in,” he says.

“That’s five minutes back in a day for a CFO per order.” Multiplied across teams, processes and departments, those efficiencies begin to add up.

“Pronto have done a great job because they’ve focused on actually getting some practical use, not just ticking a box.”

Into the future

Advances in AI-enabled ERP mean customer problems that weren’t practical or economical to solve with AI even 18 months ago are now far more realistic.

“Certainly for smaller customers, it gives access to automation that previously they would have required reasonably involved integration or automation projects to remove manual steps and change.

“Some of these AI agents and processes will enable that automation and remove that friction almost out-of-the-box.”

Longer term, the roadmap extends into agent-to-agent interactions, where systems communicate directly with customers, suppliers and other applications.

“The future vision is kind of agent-to-agent communication,” English says.

Yet despite the focus on new capabilities, he is adamant the role of ERP itself is not changing.

“ERP is still at the crux of the business. It’s still the core of business,” he says. “AI is just an augmentation to that to enable better ways of working and new ways of working.”

For organisations still trying to determine where AI belongs in their ERP strategy, English’s advice is to look beyond the hype and focus on the everyday frustrations that slow people down.

“There is a lot of hype around AI, but there are very tangible use cases where it provides  productivity enhancement. And that is real.

“We’re seeing it all over the place.”

The post ERP’s biggest AI wins might be the ‘boring’ ones appeared first on iStart keeping business informed on technology.

The proposals, contained within new consultation paper from the Department of the Prime Minister and Cabinet on building AI infrastructure that works for Australia, form part of a broader push to develop national AI standards covering both large data centres and the training of advanced AI models in Australia.

“Australia cannot simply be a destination for infrastructure investment – we must and will leverage these advantages and translate them into long-term benefits.”

While much of the government’s AI infrastructure agenda has previously focused on issues including energy demand, water consumption, community impacts and planning approvals, the consultation also proposes mechanisms aimed at ensuring AI training activity delivers broader benefits to Australia.

“Australia cannot simply be a destination for infrastructure investment – we must and will leverage these advantages and translate them into long-term benefits for the Australian economy and people,” the consultation paper says.

It says Australia is well placed to attract investment in large data centres and frontier model training but argues that investment should be developed ‘for the benefit of Australians’ and proposed conditions for AI training that could include support for local capability, skills, research and innovation, saying frontier AI developers are ‘uniquely positioned’ to contribute to local research capability, technical talent development, university, TAFE and registered training organisation partnerships, and innovation ecosystems that cannot be fostered through investment in AI infrastructure alone.

“Investing in future industries and research and innovation, will help Australia build an enduring research base, strengthen local capability and capture more of the intellectual property and commercial value arising from AI training,” the paper notes.

Without mechanisms to encourage onshore research activity and investment in the broader research and innovation ecosystem, it warns Australia risks hosting AI infrastructure and training activity while the associated research outputs, model development, intellectual property and commercial returns accrue offshore.

Among the options suggested for securing contributions are public-interest research payments and reserved access to computing resources for research, public-interest activities and negotiated in-kind contributions. The paper also asks whether frontier AI developers are best placed to make contributions for local capability, skills, research and innovation, and whether others, such as hyperscalers and neocloud providers, should also be considered.

The proposals sit alongside a range of previously flagged requirements for large data centres, including expectations around energy use, water consumption, community engagement and workforce development. Prime Minister Anthony Albanese recently backed down on plans to force new data centre builds to be fully powered by renewable energy.

The government is also seeking feedback on requirements designed to ensure data centres do not increase costs for consumers and communities and make positive contributions to Australia’s energy transition.

AI’s ‘defining influence’ on Australian economy

The consultation document comes as the Treasury releases its Intergenerational Report 2026 which identifies AI as one of the most significant forces expected to shape Australia’s economy over the next four decades.

The report – the seventh in the series – describes AI as a potentially transformative general purpose technology and says Australia is well positioned to benefit from AI adoption and investment ‘if we get things right’ because of its stable institutions, renewable energy, potential and growing role as a destination for data centre investment. It also notes that investment by hyperscale cloud providers has accelerated significantly in recent years, with global capital investment expected to exceed $1 trillion in 2026.

“In Australia, the impacts of the AI investment boom are currently most pronounced in

physical infrastructure. Current industry estimates suggest Australia’s data centre pipeline

could support more than $150 billion in investment by 2030, the equivalent of five per cent

of today’s nominal GDP.”

It notes that AI will transform Australia’s broader industrial base as new technologies lift workplace productivity and change the skills in demand and has the potential to make essential services more accessible, affordable and responsive to individual needs.

However, the timing and extent of those changes is ‘uncertain’ as it depends on the rate of technical innovation and adoption.

“The challenge for Australia is to harness and share these benefits widely whilst mitigating its potential to generate societal harms such as misinformation, scams and malicious cyber operations.”

According to the consultation paper, attracting AI training activity could strengthen national security, science, innovation, resilience and business productivity. However, it argues that investment should also help build domestic capability.

Among the questions posed in the consultation are what obligations should apply to organisations undertaking frontier AI training in Australia, how AI training could support local research and innovation, and what mechanisms could be used to strengthen Australian skills and capability development.

The paper also raises the prospect of AI training activity contributing directly to Australian researchers and institutions, including the option of public-interest research payments and reserved computing capacity that could be used by researchers and public-interest projects.

The consultation does not recommend a preferred option and is seeking industry and public feedback on whether such measures are needed and how they could operate.

Consultation is open until 9 October 2026 with Albanese previously indicating that he wants to bring legislation to parliament in early 2027.

The post Australia seeks AI dividend from infrastructure boom appeared first on iStart keeping business informed on technology.

A decade ago farmers were weighing lambs by hand, estimating pasture growth by walking paddocks with plate meters and making decisions using averages. Today, AI systems, sensors, automated weighing, satellite data and digital farm platforms are turning every animal, paddock and hectare into a source of real-time intelligence – fundamentally changing how farms are run.

“Our biggest limiting factor was ourselves and how we thought traditionally about how we’ve done things in the past.”

“Technology and AI have given me more time to spend thinking about and deciding on those higher value decisions, working on the business rather than in the business doing stuff day to day,” Pawson-Edwards said at this week’s AWS Cloud and AI Day in Auckland.

Recent research by ANZ Bank has shown agritech could provide a $2.9 billion opportunity for farms, uplifting revenue by five percent and decreasing costs by five percent.

Its report highlighted better-connected information systems, more efficient use of inputs, targeted automation, smarter land-use decision and investment in tools that genuinely fit the farm system.

For many farmers, the issue isn’t lack of information. Instead, it’s information siloed across multiple systems, from accounting systems and production records, to environmental reporting tools, spreadsheets and advisers processes.

“Farmers have a wealth of data and what they’re looking for is not more data but interpretation and decision support in that space,” says Mark McAtamney, chief digital officer at farmer owned agricultural co-operative Ravensdown.

Pawson-Edwards, who says Lone Star is only at the beginning of its tech and AI work, says at last count the business had 32 applications or programs that were all ‘keeping data in their corners’. “We’re probably only using 10-20 percent of that data to make decisions. For us, it’s going to be unlocking all that data and putting it into a format that we can make a decision at our fingertips and not be waiting a month to process it all.”

McAtamney says one important area is collaboration across suppliers and producers to ensure the data captured on the farm is able to be taken advantage of. “We’re seeing a lot of improvement in the collaboration between those businesses that are supplying or producing product from farm”

For Cheyne Gillooly, CEO of New Zealand Young Farmers, the opportunity extends well beyond individual farms.

“We’ve got some of the best sheep genetics in the world, we’ve got some of the best dairy genetics in the world and we’ve got some of the best plant and varietal information and it’s sitting in IP on shelves. Previously we would have to have scientists and entrepreneurs trawl through that for years to try and find the next chemical or next compound.

“AI can do that for us now so rapidly, so the pace at which we’re going to be able to innovate and generate new products that come from the land means we can unlock value that’s been sitting there potentially for decades.”

He says we’re moving from a world where technology and value were driven by speed and pace of innovation to one driven by ‘something that is tangible, something that is focused on human health, longevity and wellbeing, land-based products with our genetics that sit on top of that’.

“New Zealand is sitting on a good mine. The value of that data that’s currently untapped – it’s the magic of AI.”

He believes we’re on the precipice of exciting times for agriculture and farming as AI makes knowledge ubiquitous.

Pawson-Edwards admits he was sceptical when new systems were introduced. “To start with, I didn’t believe this sort of stuff was possible,” he says. “We had these people come and tell us we can lift pasture growth by this, we can improve that.”

The farm began cautiously, fitting 500 Halter digital collars across a herd of 2,500 cattle. The results came quickly. “Within probably three weeks it had changed our whole concept of this can actually work and is going to be really powerful.”

The farm now has pasture covers straight to the phone daily, rather than waiting a month for the data previously collected through a full-day of towing a plate meter behind the quad bike.

Looking back, he says the biggest obstacle wasn’t technology.

“I think probably our biggest limiting factor was ourselves and how we though traditionally about how we’ve done things in the past.”

Trust remains central to adoption. Farmers want evidence that systems work, recommendations are accurate and the data reflects what they see on the ground.

McAtamney recalls occasions during prototyping of a Ravensdown system where machine generated recommendations appeared unusual during testing.

“We would see something really unusual come out on the recommendation,” he says. They’d put the farm map on the table in front of the farmer and tell them ‘we think this is an error’. Their response: ‘Oh no, absolutely not. That’s exactly how we treated that paddock last year.’

Those moments helped build confidence that the models were correctly identifying patterns experienced farmers already understood.

Andrew McLaren, VP product and engineering for Kiwi agritech Halter, says at least half of the features in Halter have been added thanks to input from farmers.

“Once they understand what it can do, they can see the tiers to the possibilities of the technology so there have been a lot of creative ideas over the years that have ended up in the product.”

For all the attention on AI, the technology’s success in agriculture may ultimately be measured by something much simpler: whether it helps farmers make better decisions. If that happens, New Zealand’s next competitive advantage won’t be bigger farms or more animals. It will be the ability to extract more value from every hectare, every paddock and every data point.

The post AI on the farm: From mob management to precision farming appeared first on iStart keeping business informed on technology.

AppWrap September 2026

24.09 An OpenAI developed AI agent accessed public and ‘non-public’ Medicare files in June after ‘infiltrating’ the public facing Medicare statistics reporting service portal, administered by Services Australia, Prime Minister Anthony Albanese says. A forensic examination is now underway to determine more details, including what other government systems were affected Albanese says. The portal contains ‘non-sensitive’ information relating to data and statistics such as spending, with no personal information believed to have been accessed ‘at this stage’. Albanese says he had a ‘very frank’ discussion with OpenAI CEO Sam Altman and says the company too ‘far too long’ to inform the government what had happened. “The nature of the way that notification occurred as well was unacceptable,” Albanese says.

23.09 Sydney and Melbourne councils have called for federal and state governments to enact moratoriums on data centre development until planned national standards are implemented. InformationAge reports local councillors attending a Senate committee hearing expressed concerns over accelerating developments in their jurisdictions.

22.09 AI data centre startup Firmus is gearing up for a $7 billion IPO next month, with a likely target valuation of $50b for its ASX IPO, StartupDaily reports. The $50b would put it second only to Telstra’s 1997 float and ahead of Canva. Firmus was valued at $15b last month after raising US$2b for its Project Southgate, a $73b national data centre rollout.

18.09 Canterbury-Bankstown Council has voted to ban smart glasses from its leisure centres and pools, with the ban allowing staff to remove people with smart glasses from facilities, ABCnews reports.

16.09 Claude developer Anthropic has signed a lease for the site of its first $31.9 billion data centre in Queensland. The project, which is yet to be approved by the local council, is expected to be built in four stages with its first stage planned to open next year, the Guardian reports. Queensland premier David Crisafulli has heralded the deal as a major win, saying it will ‘deliver more jobs and put more energy into Queensland’s grid’.

16.09 Dozens of approved but yet-to-be-built AI data centres will escape the federal government’s proposed restrictions around energy and water use because new laws are not expected to be retrospective, ABCnews reports. The combined capacity of 27 approved, but unbuilt, data centres is at least 3.38 gigawatts.

15.09 A Donald Trump appointed prosecutor has asked a judge to overturn several findings ruling an Atlassian engineer was unfairly dismissed, the AFR reports. The move would give Atlassian and other employers more control over workplace speech and employee exit agreements. The company had been found to have broken the law in firing the engineer and was ordered in July to reinstate her with back pay. Atlassian appealed the judgement and general counsel Cystal Sowe Carey has asked the administrative law judge to reverse several findings, but maintains the worker was illegally dismissed.

09.09 CEDA’s latest business insights report shows data centres, cloud computing and AI infrastructure have propelled Australian investment to its highest GDP share since the mining boom a decade ago. InvestorDaily reports business investment reached 12.6 percent of GDP in the June quarter and is expected to peak at more than 13 percent during 2027-28. Business investment in software also grew up nine percent year on year. One analyst however warns the data centre investment boom is not an economic saviour with ‘negligible’ likelihood that the current data centre build out will result in material uplift in near term economic growth.

09.09 Western Australia has launched a review of government procurement rules aimed at reducing red tape and improving access to government contracts for smaller businesses.

08.09 Draft legislation would enable users to opt out of algorithms on social media platforms, enabling them to only see content from individuals and groups they have chosen to follow, rather than divisive content often pushed by the algorithms, The Guardian reports. The new laws would also require services such as online games, apps and AI chatbots to protect children from specific harms, with the eSafety commissioner having the power to issue removal notices for content. The plans face significant opposition in federal parliament.

08.09 Non-profit security organisation Shadowserver Foundation says 382 Exchange servers in Australia remained exposed to CVE-2026-62911 as of August 31, three weeks after Microsoft issued a fix. The Australian Signals Directorate has issued warnings urging organisations to running the outdated Exchange versions to take action or segment their networks if they cannot replace them, TechBest reports.

07.09 The personal data of more than one million students, parents, guardians and school staff in Australia and New Zealand has been exposed through an attack on online learning platform Mathspace. 7News reports names, email addresses and user IDs were among the information affected, along with Mathspace staff records. Mathspace says customer passwords, single sign-on tokens and other customer authentication credentials were not exposed.

03.09 Airservices Australia is exploring an ‘AI front door’ broker platform within its AWS tenancy to act as a single governed entry point for all AI interactions, ITnews reports. Airservices hopes the platform, to be trialled by around 100 users before being expanded to 3000+, will stop the proliferation of unmanaged point solutions.

AppWrap August 2026

31.08 The second tranche of privacy act amendments includes a proposed 72 hour deadline for notifying the OAIC of eligible data breaches, together with a positive obligation to maintain breach-response practices and take reasonable steps to contain harm, Dentons says.

27.08 The federal government has backed down on requiring data centres to rely on renewables with new standards, set to be legislated next year, leaving room for some jurisdictions to use fossil fuels as well, TheConversation says.

27.08 Australia’s financial regulators APRA and ASIC are urging organisations to move from mere awareness to taking decisive action around frontier AI risks. They says boards and executives must move beyond awareness and ensure their organisations have well-tested response plans and understand where they are vulnerable so they can respond effectively under pressure. The regulators flagged a growing interest in defensive AI, including for threat intelligence, vulnerability detection, code review and incident response along with the importance of getting cyber fundamentals right.

27.08 Woolworths has revealed an AI employee support tool is handling around 7,000 staff enquires a week and resolving nine out of 10 of them without human intervention. ITnews reports the retailer has also used AI to cut weekly catalogue production from a week to a few hours.

27.08 Xero investors have voted against adopting the board’s decision to give CEO Sukhinder Singh Cassidy a 22 percent pay rise – lifting her target remuneration to US$18 million, the majority through long-term equity awards split between performance and service, Forbes Australia reports.  The move comes amid ongoing struggles for Xero’s stock prices, which have been halved in the past year. Under NZ law the board does not require shareholder approval, but put it forward to an advisory vote as ‘a matter of good governance’.

27.08 Meta has agreed to a landmark US$18 billion settlement in a US federal case accusing it of endangering children, with a slew of new safety features now required to its platforms including Instagram and Facebook. Al Jazeera reports those changes range from night curfews to two-hour usage limits for US users under the age of 18. Other countries are also taking regulatory action against Meta and other social media platforms.

25.08 The Australian Energy Market Operator and Oxford Economics Australia are forecasting a seven-fold increase in data centre electricity demand in Australia over the next decade as AI and cloud infrastructure investment accelerates. They say under the central scenario data centre electricity consumption across the NEM is forecast to increase from 5.1TWh in FY26 to 15.5 TWh by FY30 and 33.7 TWh by FY36.

19.08 The ACCC has raided the offices of logistics software company WiseTech Global for documents and electronic data related to the supply of global logistics services and software. The search is part of an investigation into alleged contraventions of the Competition and Consumer Act 2010, WiseTech says, adding that it intends to fully cooperate with the investigation.

19.08 Quest Apartment Hotels says it is investigating a security breach which has compromised customers’ personal data relating to records before June 2025 and including full names, email addresses and contact details. Some data entries also include customers’ date of birth. Quest says unauthorised access to a database system arose from a vulnerability through a third-party service provider, ABCnews reports. The OAIC and Australian Cyber Security Centre have been notified.

18.09 The Origin Energy cyber hack which exposed the personal data of nearly a million Australians has been traced to a former Accenture employee in a Manila call centre, ABCnews reports. Information of around 900,000 current and former customers was accessed in the breach, which was confirmed last month.

11.08 Bendigo and Adelaide Bank has received an $8m fine after admitting numerous cybersecurity failures in relation to a 2023 cyber attack on its former Alliance Bank business. The attack saw a hacker access 257 customer accounts and extract funds. APRA says there were ‘significant weaknesses’ in customer authentication controls for online banking, including password settings that allowed very weak passwords, and system design features that enabled the treat actor to identify valid customer IDs. A number of the weaknesses had been identified in 2020 but not addressed, BusinessNews Australia reports.

12.08 Social media platform X (has won a lawsuit it brought against Australia’s eSafety Commissioner after the Federal Court found the company does not need to follow a new safety standard for harmful online content. InformationAge reports X successfully argued it could not be designated under Australia’s Online Safety Act as both a ‘social media service’ and a ‘relevant electronic service’ (RES). The RES standard therefore ‘does not apply’.

07.08 Australian AI infrastructure company Firmus has raised US$2 billion in equity in its latest funding round as it looks to speed up build-out of AI factories in Australia and Asia Pacific. Reuters reports the raise brings the company’s post-money valuation to more than $10.5b.

AppWrap July 2026

30.07 Total capital raise by Australian cyber startups has bounced back to $171 million in the 2026 financial year following a 2025 lull. Major rounds like a $105 million raise at risk monitoring company Upguard and $30 million at Kasada headlined the rally, AFR says.

30.07 Elon Musk’s X has accused Australia of trying to give eSafety unfair information-gathering powers and undermining US law with its plans for tougher enforcement of the under-16s social media ban, Reuters reports.

30.07 Despite a 115 percent year-on-year surge in local AI spending, only 23 percent of Australian companies have auditing processes to verify AI outputs according to research by ServiceNow. Just 17 percent have a system to manage and track governance and compliance, the vendor says.

29.07 Queensland and the Northern Territory have rejected federal government plans to mandate that AI data centres use renewable power, The Guardian reports. Their refusal comes as S&P Global warns power use from data centres could risk five-fold by 2035 to 10 percent of Australia’s total consumption and lead to energy bills skyrocketing.

20.07 The Australian Computer Society has called for a national framework to address Australia’s cyber skills shortage, following new research showing unclear pathways, inconsistent job definitions and high barriers to entry are holding back the cyber security workforce, CyberDaily reports.

15.07 Environmental and community groups are calling for a pause on data centre developments until new regulations are in place requiring data centres to underwrite new power supply, pay their full share of grid connection and add as much energy to the grid as they take out of it, The Guardian reports.

15.07 The Australian government is creating an Office of AI and introducing standards requiring large AI data centres to underwrite new power generation and limit water use. Reuters reports legislation will be introduced in Parliament early next year. The new office will sit within the Department of Prime Minister and Cabinet.

13.07 Digital customs declaration form will be phased in at all Australian international airports and seaports over the next 18 months following a successful trial with Qantas. The federal government says it will spend $56.1m over four years to modernise airport systems and introduce the digital travel declaration, the Guardian reports.

11.07 Meta has abruptly pulled a controversial AI tool which enabled people to create fake images from user content on Instagram. The feature, part of the Muse Image rollout, quickly sparked blowback due to privacy concerns as anyone with a public account could have their likeness used without their permission or knowledge, the BBC reports.

10.07 The Commonwealth Bank of Australia is planning to cut a further 176 tech and engineering jobs, according to the Financial Sector Union. InformationAge reports the FSU is accusing CBA of ‘fake redundancies’, by advertising similar roles under its Indian arm. CBA has denied the claims.

07.07 A hacker is selling what they claim is live Teletrac Navman telemetry data from almost 3,000 Australian and New Zealand organisations. CyberDaily reports the hacker claims to have exfiltrated continuous real-time GPS tracking data for 2,988 customer companies across A/NZ, over a 48 hour period, with the data including emails and mobile phone details for drivers, driver’s licenses and vehicle registration details.

07.07 Australia’s newly established AI Safety Institute is already examining some of the world’s most powerful AI models, according to assistant technology minister Andrew Charlton. The work forms part of efforts to improve understanding of emerging AI risks and capabilities.

07.07 Most Australian’s don’t want to live near an AI data centre according to a survey from Primara research. CyberDaily reports 83 percent of the 1000 surveyed would oppose having an AI data centre in their local area, with only 16 percent believing nearby communities would receive meaningful benefits from such developments.

07.07 The Queensland Government has announced a new digital investment and assurance process which will review and monitor all new digital and ICT investments over $5m, including projects supported by the $1b Queensland Government Digital Fund. The Minister for Customer Services and Open Data, Steve Minnikin, says more than $300m is being invested in ‘high-impact digital initiatives’ to modernise critical government systems and improve service delivery.

07.07 ANZ Banking Group has outlined a new technology roadmap under CIO Gerard Florian, signalling a strategic shift in how the bank approaches technology modernisation and digital delivery. ITnews reports the move comes six months into Florian’s tenure.

03.07 Microsoft has signed a deal to pay to use news content from Nine in its Copilot AI search results in a first-of-its-kind deal in Australia. The deal, enabling Copilot to reference text of Nine’s content, beyond paywalled previews during AI searches to ‘contextualise and ground’ outputs, covers content from mastheads including The AFR, Sydney Morning Herald, The Age and Brisbane Times, Microsoft says.

02.07 Australia has regained access to Anthropic’s Fable AI, but not to Mythos as the US Department of Commerce reverses its export ban on the models. InformationAge reports Fable has been redeployed for public use, but Mythos has only been restored for a specific set of US organisations.

02.07 University of Sydney academic Olivia Shen has warned Australia shouldn’t try to build its own frontier AI. Writing in The Conversation, Shen, from the US Studies Centre, says the eye-watering cost some countries have invested in training model, hasn’t yet produced AI models with comparable performance to leading US or Chinese offerings, and says Australia needs to make pragmatic investments where we have greater maturity and agency.

AppWrap June 2026

30.06 Bendigo Bank is building what it believes is Australia’s first agentic SOC. The Bank’s CSO says security operations are set to be increasingly supported and underpinned by agentic AI capabilities, with internal security teams focused on more advanced high-value work, ITnews reports.

23.06 The Australian Signals Directorate has joined other Five Eyes partners in warning government and corporate leaders that swift action is required to remain ahead of the increased risk from AI. The agencies are warning leaders to understand and assess risk, readiness and accountability, prioritise foundational cyber security practices and controls, empower cyber leaders with authority and resources and stay actively engaged. Their warning includes practical actions which while not new are now deemed ‘urgent’ by the agencies including reducing attack surfaces, accelerating patching and limiting who can access critical systems.

23.06 More than 4,000 Tasmanians have signed petitions opposing AI data centres, including 3,500 opposing Firmus Technologies planned Bell Bay facility, which will use up to 288 megawatts of power. PulseTasmania reports the facility will be the state’s second largest power user behind the aluminium smelter which draws 335 megawatts and employs more than 550 people. Firmus is also building a $2b data centre near Launceston, with a third flagged for Wesley Vale.

19.06 KPMG has been grilled by a parliamentary committee investigating whistleblower claims around the misuse of confidential client information to bid for new business. ABCnews reports senior KPMG execs fronted for the inquiry which heard about the alleged misuse of confidential information belonging to client Lendlease, its treatment of the whistleblower and a third KPMG confidentially breach involving client Optus.

18.06 The NSW government’s procurement board has issued a sweeping moratorium for all government agencies and departments on any new procurements involving KPMG. The ban comes amid ‘serious concerns’ about KPMG’s work practices. The company is embroiled in a series of allegations about misuse of confidential client information to bid for new business. An approval and reporting process for KPMG engagements has also been established in NSW.

18.06 The federal Department of Parliamentary Services has failed key cybersecurity benchmarks with protections still only ‘partly effective’ seven years after the network was breached and seven out of eight key cybersecurity controls falling short, InformationAge reports.

18.06 PsiQuantum has broken ground and started construction of its billion-dollar supercomputer facility in Melbourne. The company says it will build and deploy the world’s first utility-scale, fault-tolerant quantum computer at the Moreton Bay Central facility. It has received nearly $1b in federal and Queensland funding for the supercomputer.

17.06 The NSW Government has denied claims by ransomware group Nova that it has been breached, saying sample files provided are publicly available and historical information. CyberDaily reports Nova listed the state government as a victim on its darknet leak site, claiming to have exfiltrated sensitive data from the network.

17.06 A new Microsoft Teams feature enabling Teams to detect activity on a user’s device even when they are not actively using Teams has raised privacy concerns. Auckland University lecturer Alex Baird warns companies that the Privacy Act is clear on being thoughtful about why data is collected and using it only for that purpose. If used for checking teammates know where you are, that’s great he told RNZ. Collecting data for performance tracking however would be ‘a very different thing’.

16.06 Anthropic has disabled its new Claude Fable 5 and Mythos 5 models after a US Export Control order. Forbes reports the US Commerce Department mandated the suspension of access for all foreign nationals citing security and export control powers. The move followed reports of a jailbreak bypassing safety guardrails.

12.06 SpaceX has made the biggest stock market debut in history, making Elon Musk the world’s first trillionaire. Early trading saw share prices jump from the opening $150 a share by a double digit percentage, sending the company’s valuation above US$2tn, the Guardian reports.

11.07 Woolworths is moving IT jobs offshore to ‘remove complexity and drive efficiency’. Hundreds of jobs across the people, IT and finance teams are expected to be affected, ABCnews reports.

10.07 Tasmania’s Aurora Energy has kicked off an ERP modernisation program, replacing legacy systems with Microsoft Dynamics 365 finance, supply chain management and project operations, ITnews reports.

09.07 Anthropic has released a new version of its Claude Mythos AI model to the general public, but is restricting its use in sensitive areas. The Guardian reports Fable 5 is the first model to be made widely available from the company’s new Mythos class, unveiled in April but restricted to a small set of users over cybersecurity concerns. Anthropic says Fable 5 will be useful for writing and debugging software code, answering complex research questions and analysing images.

09.07 Perth Airport is deploying more than 70 IT and OT systems for its new terminal development. DXC, which has been selected as the master systems integrator for the program, says it includes a broad range of tech solutions including AI, cloud and cybersecurity across the new terminal for passenger processing systems, baggage tracking and reconciliation, advanced security screening integration, building management and automation, digital signage and wayfinding, public wifi and operational control systems.

04.06 Meta has accused Australia of violating a free trade agreement with the US by proposing a new 2.25% tax on tech giants who don’t have licensing deals with local media. The company claims the tax is broader than existing digital services taxes which have prompted US trade actions. Reuters reports a spokesperson for the Assistant Treasurer says the government remains committed to the change.

04.06 Australia ranks equal lowest out of 23 countries when it comes to AI sentiment according to an EY survey, which found while 51 percent of Australians like the idea of using AI to simplify life and support choices, 68 percent are worried about losing control over decisions made by AI on their behalf. ABCnews reports just one percent of Australians say they have complete trust that AI will be used responsibly.

03.06 The Department of Defence is using the Maven smart system from controversial company Palantir but says it is in a sandboxed environment and without AI functionality, ITnews reports. Defence has two active contracts with Palantir, including a $10.4m deal signed in February. It says it is using Maven to aggregate data ‘to inform targeting solutions’.

02.06 Florida has become the first US state to sue OpenAI, over the design and safety of its ChatGPT chatbot. The state is suing both OpenAI and CEO Sam Altman, alleging they know ChatGPT is not safe, especially for minors, CNN reports.

AppWrap May 2026

27.05 Today’s students may have grown up with tech, but they’ve just recorded their worst-ever results in national digital literacy tests, ABCnews reports. Just 37 percent of year 10 students and 50 percent of year 6 students were accessed as proficient – the lowest level for year 10s since testing began in 2005.

27.05 The Federal Court of Australia has called in CyberCX to test data and security controls of the platform used by VIQ Solutions court transcription service. The work follows February’s revelation that some transcription work had been subcontracted to an Indian company, in breach of contracts, and confirmation that there had been data privacy incidents in the performance of the services, ITnews reports.

22.05 Australian quantum computing startup Diraq has signed a $53m grant agreement with the US government to establish a supply chain in the US. Forbes Australia reports much or all of the grant funding will turn into equity making the US government an investor. Diraq is also eyeing up a US Defence deal worth up to $420m.

15.05 The Federal Budget 2026 is a mixed bag for tech companies with the end of capital gains tax discounts expected to have a major impact on founders of tech companies, while an R&D tax incentive and loss refundability changes could be a powerful funding strategy for startups, accounting and advisory services company WilliamBuck says. It also applauded the expansion of VC tax incentives.

13.05 New tax breaks for innovative companies, tax deductions for workers, more funding for CSIRO, an overhaul of the migration points test and changes to capital gains taxes all feature in this year’s Federal Budget. Up to $70m in AI Accelerator grants, tens of millions of dollars to expand AI use in government departments, $654m+ over four years, and $166m/year after that, to expand the Digital ID system and $357m for the ATO’s myID mobile app were among the funding allocated InformationAge reports. The CDR will also see expansion with an additional $62m allocated over two years, while cybersecurity, defence and space also saw boosts.

13.05 The parent company of the Canvas educational tool involved in a major global hack last week has struck a deal with the hackers to secure stolen student and school data, Reuters reports. All data was returned to the company, it says, and digital confirmation of data destruction along with a promise no Instructure customers will be extorted were received Instructure says.

12.05 Xero CEO Sukhinder Singh Cassidy has apologised to customers after the accounting platform was hit with five days of outages. Xero says it has contacted the ATO about deadlines which will be affected by the outages with the ATO ‘understanding’ of the situation, SmartCompany reports.

07.05 Legal practice management platform Smokeball has taken top spot as the best medium to large tech workplace in Australia’s Best Workplaces in Technology, followed by consulting and engineering company Mantel Group and cybersecurity company Crowdstrike. In the small category Macquarie Cloud Services took top for the second consecutive year followed by data and AI consultancy V2 AI and software company Corto, while in the micro category AI Technologies was followed by Adora and Causeis, Great Place to Work says.

07.05 Australia’s largest tech startup community, Fishburners, is in voluntary liquidation, with KPMG appointed administrator. InformationAge reports KPMG is facilitating a strategic restructuring of the organisation, and will be seeking expressions of interest from parties in the innovation and tech sectors. Fishburners will continue to trade as usual while the assessment is underway.

06.05 Australian education institutions are among those caught in a worldwide cyber security attack after they Canvas learning management system, developed by US company Instructure, was subjected to a hack. ABC News reports state schools in Queensland and Tasmania, universities in NSW and South Australia and TAFE in Tasmania are affected. Instructure says it doesn’t believe birthdates, passwords, government identifiers or financial information have been impacted. Names, locations of study, email addresses and messages between users are however believed to have been compromised.

04.05 NAB has changed its capitalised software policy, reducing the value and useful life of capitalised software assets thanks in part to AI enabling software to be built or replicated quickly and cheaply. ITnews reports the bank took a $1.3b hit to its underlying profit and $949m to its cash earnings for H1 FY2026 owing to the change.

AppWrap April 2026

28.04 Minister for Employment and Workplace Relations Amanda Rishworth says her department is currently undertaking gap analysis to identify how current frameworks and institutions are interacting with the adoption of AI. Speaking at the AFR Workforce Summit, Rishworth noted the tripartite AI Employment and Workplaces Forum, which meets for the first time this week, bringing together government, employers and unions to ‘build a common understanding’ and translate themes of trust, capability, transparency, safety and productivity into actions and outcomes in workplaces.

27.04 NAB is building an AI science team to guide how AI tools are deployed across the business and the workforce’s transition. TheAussieCorporate says the team will shape which tasks are automated and which new roles are created to support them. The work includes designing systems to safely handle customer data, helping make staff decisions and streamlining back office processes.

23.04 The Australian government has signed a MoU with Microsoft which is promising to continue investment in local AI capability and cloud computing and align with the governments Expectations for Data Centres and AI Infrastructure Developers. The Department of Industry, Science and Resources says the MoU – the second under the National AI Plan – also outlines Microsoft’s intention to collaborate with government on AI safety and workforce capability, support delivery of the APS AI plan and help the government understand future infrastructure needs.

22.04 The OAIC has found that InspectRealEstate’s rent-tech platform, 2Apply, collected excessive information and did so by unfair means. The OAIC says the platform has agreed to adapt its personal information collection practices on a without-admissions basis. The decision requires IRE to cease collecting personal information just as prospective renters’ gender, student status, citizen status and visa expiry and details of previous living history.

21.04 Identity verification company Persona has announced an integration with ConnectID, the Australian digital identity exchange created by Australian Payments Plus. BiometricUpdate says the integration keeps data private, providing age verification that relies on data a financial institution has already collected from a user.

22.04 The eSafety commissioner has given legally enforceable transparency notices to Roblox, Minecraft, Fortnite and Stream amid concerns online games are being used to spread violent propaganda and radicalise young people and by sexual predators to groom children. eSafety says the notices require the providers to explain how they are identifying, preventing and responding to these harms, along with bullying and online hate.

21.04 A NSW Treasury staff member has been charged after allegedly downloading more than 5,600 government documents containing ‘confidential commercial and financial information’. NSW Government says the information covers multiple departments and projects. Police believe all the data has been located and is now secure, with no external compromise to the Treasury’s system.

16.04 Apple has granted Australian law enforcement access to user notification data for the first time. InformationAge reports a transparency report shows four push token identifiers were sought across two of the requests, with one push token request granted.

16.04 South32 Aluminium is suing Siemens alleging missing code in a programmable logic controller led to a steam turbine generator overheating and being ‘effectively destroyed’, ITnews reports.

15.04 Travel giant Booking.com has notified an unknown number of customers about a data breach which has seen hackers steal customer data. BBC reports some customers have contacted it saying they have already started receiving suspicious messages. Booking.com has declined to say how many people are affected or in what regions.

09.04 Canva has acquired Australian AI tool Simtheory and Australian marketing automation company Ortto for undisclosed sums. Canva says the deals will take it from a design tool to an ‘end-to-end’ work system and strengthen its AI capabilities, 9news reports.

09.04 Bendigo and Adelaide Bank will cut its tech and business operations teams after signing two major tech deals with Infosys and Genpact outsourcing some tech and business management capability. The cuts will save the business at least $65 million yoy, by the 2028 financial year AFR reports. It adds hundreds of roles are likely to be affected.

03.04 Artemis II astronauts experienced Microsoft Outlook failures shortly after launch, with Mission Control called in to remotely troubleshoot the issue. Commander Reid Wiseman reported that two instances of Outlook were running simultaneously on his Surface Pro, leaving both unresponsive, Mashable reports.

02.04 The ATO has launched an in-app security feature enabling users to confirm, in real-time, that they are speaking with the real ATO not a fraudster. Almost 7,500 ATO impersonation scams were reported in July 2025 alone, with impersonation scams peaking during tax time, the ATO says.

01.04 SpaceX has filed confidentially for an IPO according to reports, which say the company is committed to debuting in June, with Elon Musk aiming to raise US$50b-$75b. The NYTimes reports SpaceX values itself at more than $1 trillion and would be one of the most valuable companies to reach the stock market.

01.04 Oracle is cutting thousands of jobs from its 162,000-strong workforce, with 10,000 believed to have lost their job so far. Senior engineers, architects, operations leaders, program managers and technical specialists are among those affected, the BBC reports. Around 10,000 people are believed to have lost their jobs so far. It is unknown if the cuts are related to Oracle’s heavy AI spend.

01.04 The Federal government has signed a new MoU with Anthropic to build on the national AI plan launched in late 2025.

AppWrap March 2026

31.03 eSafety has flagged Facebook, Instagram, Snapchat, TikTok and YouTube for compliance issues with the Australian Social Media Minimum Age obligation and is gathering evidence to inform potential enforcement action, it says.

26.03 Meta and Google have lost a landmark US case, with a Los Angeles jury finding the two companies negligent for designing social media platforms that are harmful to young people. Reuters reports the jury found Meta liable for damages of US$4.2 million, with Google liable for $1.8m in the case of a US woman who sued Meta and YouTube over her childhood addiction to social media.

25.03 Delaying public disclosure of serious cyberattacks on critical infrastructure operators ‘to prevent disclosure from compromising national security’ is one of five changes proposed in a Department of Home Affairs consultation paper for amended Security of Critical Infrastructure (Soci) rules. Other reforms in the paper include changes to make it easier for the government to restrict multiple entities from using high-risk vendors or technology and clarity around the threshold for issuing a direction.

25.03 Canva has acquired Melbourne-based digital out of home advertising company Doohly in a $30m deal which adds outdoor advertising to its offering. The deal is the design company’s third acquisition this year and comes just a month after it bought US startup Mango.AI and the UK’s Cavalry notes StartupDaily.

24.03 Silicon Quantum Computing has secured $20m from the National Reconstruction Fund to support scaling of its quantum processing units and Watermelon machine learning system, SmartCompany reports.

19.03 The X Money payments service, Elon Musk’s attempt to build a ‘everything app’ will enter early public access in April. Musk’s ambitions for the embedding payments into X are broader than most Western tech companies, notes PaymentsIndustryIntelligence, with Musk talking about a system that could eventually encompass savings, payments, securities and other financial activity, reducing reliance on traditional banking channels. At launch X Money is expected to offer core wallet and payment functions, including the ability to move funds within the platform.

18.03 An attempt by the CEO of US videogame publisher Unknown Worlds to wriggle out of paying a US$250m performance has suffered a setback after the court documents showed Changhan Kim asked ChatGPT how to avoid paying the bonus. InformationAge reports Kim engaged the chatbot to draft a corporate takeover strategy, which involved stymieing the release of the highly anticipated Subnautica 2. Following ‘guidance’ from ChatGPT he also posted public ‘critical messages’ on the company website.

17.03 Google has reportedly paused plans for a $20b AI and data centre hub, warning the federal government that high taxes could cause the country to miss out on investment. DatacentreDynamics reports the company has told government it is concerned that if it set up such a hub in Australia, the ATO would consider it a ‘permanent establishment’ exposing it to a 30 percent corporate tax rate. Google is currently exploring where to establish the major APAC hub.

16.03 AI disruption is causing Australian software companies to make big job cuts, with private tech companies expected to follow in the footsteps of Atlassian and WiseTech with redundancy rounds, valuation cuts and a bottleneck for public listings, AFR reports.

12.03 Atlassian is cutting around 1,600 jobs – or 10 percent of its workforce. The company says the move is to ‘self-fund further investment in AI and enterprise sales’, while strengthening its financial profile, Forbes reports. The retrenchment will cost between $225m-$236m.

11.03 Professional body CA ANZ says it will investigate the failure of an external exam delivery platform after more than 1,300 students were locked out of one exam and another 60 experienced delays of up to 40 minutes on a second exam. CA ANZ blamed a failure in an external exam platform operated by a third-party provider.

11.03 US medtech giant Stryker has been hit by a major cyberattack, with an Iran-linked hacking group claiming responsibility, saying it is in retaliation for the killing of more than 170 people – mainly schoolgirls – in a strike on a school, and warning it marks the beginning of a new chapter in cyber warfare, Al Jazeera reports. The group says it has seized 50TB of Stryker data in the attack which it says has erased data from 200,000 devices.

10.03 Anthropic is opening a Sydney office and hiring a local team. The company says Australia ranks fourth globally in Claude.ai usage, relative to population, with strong demand from local business. It says it plans to deepen engagement with Australian institutions and collaborate on projects that advance the country’s national interests and priority sectors. It is also exploring opportunities to expand compute capacity in Australia ‘given our longstanding belief that democracies should lead in AI development’. The company is currently suing the US government over its claims that Anthropic is a ‘supply chain risk’. The row erupted after Anthropic refused to all US military to have unfettered use to its AI tools.

05.03 Privacy Commissioner Carly Kind is warning Australian organisations that a recent ruling allowing Bunnings to use facial recognition is not a green light for the technology. Kind, who says she has not filed an appeal of the Administrative Review Tribunal finding, which overturned her ruling against the technology, says the decision shows the law allows for the balancing of competing interests – privacy vs public safety – but entities will need to conduct a detailed risk assessment specific to their circumstances before deploying the technology. “Retailers should view the decision as a useful case study, rather than a green light for deployment of biometric technologies,” she says.

AppWrap February 2026

27.02 Commonwealth Bank has called in the police and corporate regulator over concerns $1b in home loans were obtained fraudulently, including with AI-created documents. AFR reports the bank reported itself to police and the regulator after a review of the compliance practices and customer lending documents, which stepped up following revelations about the Penthouse Syndicate, which allegedly defrauded NAB of around $150m. Increasingly sophisticated AI has made it easier to create authentic looking false documents and to steal identities.

26.02 Nvidia has posted blockbuster quarterly results of US$68.1 billion – up 73 percent year on year and well above analysts’ forecasts. Revenue from the data centre division, which sells the chips used to train and run AI models was up 75 percent yoy to $62.3 billion, the company says.

25.02 Australian logistics software company WiseTech is cutting around 2,000 jobs – nearly a third of its global workforce in a restructure focused on using genAI to increase efficiency in software engineering and support, Reuters reports.

25.02 Australian Peter Williams has been jailed for seven years for selling critical software and information related to cybersecurity to a broker with ties to the Russian government. Williams was working in the US. 9News reports the tools were cyber-exploit components which can be used to identify weaknesses in tech systems or infect them with viruses.

24.02 The federal government has scrapped a permanent AI advisory board appointed to develop AI guardrails. ABCnews reports the body, funded in the 2024 Budget, was scrapped only months after spending $188,000 and more than a year whittling down a list of experts to a shortlist of 12 nominees. The government says it will instead establish an AI safety institute early this year.

25.02 Commonwealth Bank is spending $90 million to get staff ‘AI-ready’. The bank says the program will help employees ‘build skills, find new opportunities and get ahead of the changing nature of work’. AI learning has already been delivered to 30,000+ staff.

24.02 The ASD has publicly released its Azul open-source malware analysis tool to help safely handle and analyse malware. Azul is designed for large-scale malware analysis and is designed to be highly scalable and store tens of millions of samples, ASD says, adding it can turn common analysis steps into analysis plugins which can be used as part of an automated workflow and assist in identifying variants of a malware family more efficiently.

18.02 Xero is restructuring with 250 jobs on the line – including in Australia. The company is creating around 280 roles in Canada as part of the restructure, BusinessDesk reports [paywalled].

18.02 YouTube suffered an outage from around 11.50am AEDT, with a spokesperson for Google saying an issue with the recommendations system prevented videos appearing.

18.02 The federal opposition has unveiled a new-look shadow ministry under Angus Taylor, with Victorian senator Sarah Henderson appointed shadow communications and digital safety minister. ABCnews reports Casey MP Aaron Violi has been promoted to the outer shadow ministry with a broad portfolio including science, technology, cyber security and the digital economy. Deputy Liberal Leader Jane Hume will be shadow minister for employment and industrial relations as well as productivity and deregulation.

18.02 Australian company Terram Astra has set its sights on a US$10m seed capital raise for a new sovereign ground-based infrastructure platform designed to strengthen communications resilience, defence readiness and space safety across the region. Manufacturers’ Monthly reports Terram Astra’s first hub is located near Alice Springs.

17.02 Highly sensitive Australian court files have been accessed by a foreign entity based in India, an ABCnews investigation has revealed, with Greens senator David Shoebridge saying access to the cases by foreign entities is ‘a national security risk’. The investigation showed Canada’s VIQ Solutions, which employs transcribers in Australia, has also subcontracted work to an Indian company specialising in automated voice-to-text technology, in breach of the VIQ contract. Thousands of court files were accessed by e24 staff with Indian email addresses.

16.02 Federal government entities are failing to report cyber incidents to the Australian Signals Directorate. InformationAge reports just 35 percent of the 200 government entities said they reported at least half of all cybersecurity incidents observed on their networks.

11.02 Service NSW has started testing a new system enabling residents to verify who they are in a single step when interacting with state government services online. ITnews reports initially the digital ID system will be used with a limited number of services, starting with a state toll rebate scheme.

06.02 Australian-founded AI data centre startup Firmus Technologies has secured another $100m to expand its local data centres. The funding, from ASX-listed Maas Group means the company, which is Singapore-based, has raised more than $900m in less than five months, StartupDaily reports.

06.02 CommBank says it has become the first bank to disclose how it is ideating, developing, deploying and managing AI at an organisational level, including ‘practical examples’ of AI in action across the business. Those examples include AI to protect against fraud and scams, strengthen cyber security, enhance customer experience and detect abuse in transaction descriptions.

05.02 Atlassian’s revenue for the quarter to December 31 was up 23 percent to US$1.58 billion, with cloud revenue growing 26 percent providing the company with its first $1b cloud revenue quarter. However, net loss also grew from $38.2m to $42.6m. Operating loss was down from $57.5m to $47.7m the company says. CEO Mike Cannon-Brookes dubbed the quarter ‘fantastic’, adding: “We’re building a bloody great business.”

03.02 The federal government has rejected a bipartisan proposal which would have allowed a parliamentary committee to cancel large consultancy contracts, saying its own reforms are doing the job. A centralised register for conflicts of interest breaches has also been ruled out.

03.02 Australia’s STEM capability is facing a ‘severe crisis’ as a result of flatlining Commonwealth funded research, Science and Technology Australia says. In a pre-budget submission it says Stem research must stop being viewed as a cost and instead be seen as an investment. It says 47 percent of Stem professionals are considering leaving their current roles, with 33 percent planning to leave the sector altogether.

03.02 Xero says more than two million subscribers are using its full AI features, with more than 300,000 using its newer GenAI features. In a market update today the company claimed customers are saving around 22 hours a month using bank feeds and automated actions, with more than 12 percent using AI Insights.

03.02 Kiwi AI startup Teacher’s Buddy has raised $2.3m in trans-Tasman seed funding. The offering aims to reduce teacher workloads, helping with marking and student report writing and producing customised, differentiated curriculum-aligned teaching and assessment materials  The company says the funding was led by Auckland’s Soul Capital and Australia’s Giant Leap.

02.02 Victoria has set its sights on being Australia’s AI capital with the launch of an AI Mission Statement setting out the Allan government’s plans across six key pillars of investment attraction and adoption, data centres and digital infrastructure, local innovation, products and services, talent and workforce, ethical AI use and public sector adoption. The plan includes a $8.1 million Digital Jobs – AI Career Conversion program to safeguard jobs in industries at risk from AI, upskilling workers to transition into AI roles and become specialists in the technology.

AppWrap January 2026

27.01 Airwallex boss Jack Zhang says the fintech is at least two years away from a public listing, despite regular references to an IPO in 2026. Zhang’s comments, reported in StartupDaily, follow news last week that Austrac has ordered an audit of the company over AML/CTF concerns.

26.01 The APS plans a whole-of-government learning technology ecosystem following a co-design involving teams from 37 Commonwealth agencies. ITnews reports a core goal of the new system is to enable interoperability and integration across learning and development environments.

23.01 TikTok has finalised agreements with backers including Oracle, Silver Lake and Emirati company MGX to establish a US joint venture. Each of the three backers will hold a 15 percent stake in the company. ByteDance keeps a 19.9 percent share. The BBC reports the content recommendation algorithm has been licensed to Oracle – headed by Trump ally Larry Ellison – which already oversees US user data under a previous arrangement set up over security concerns. The deal will enable TikTok to continue operating in the US, but is likely to continue to be scrutinised, with some Democrats voicing concerns about the ties between Trump and TikTok’s new investor group could limit what gets shared on the platform.

22.01 Australia’s financial intelligence unit, Austrac has ordered an external auditor be appointed to assess whether fintech – and unicorn – Airwallex is complying with anti-money laundering and counter terrorism financing. Austrac says it has ‘concerns’ about potential non-compliance including that Airwallex’s transaction monitoring program has not been ‘attuned to the full range of risks it faces’ and the company hasn’t demonstrated an acceptable understanding of who its customers are and what reporting may be required. The auditor must report findings within 180 days of appointment.

20.91 AI projects to improve construction safety, enhance space safety and detect speech delays, are among the 174 projects winning funding in the latest round of Australia’s Economic Accelerator (AEA) Ignite program. More than $72.5m in funding was allocated. The funding is for projects that build capability I nationally important sectors aligned with the National Reconstruction Fund’s priority areas, AEA says.

20.01 Gilmour Space has become Australia’s first space tech unicorn after raising $217m in a Series E funding round – its largest funding round to date. SpaceNews reports the round was led by the federal government’s National Reconstruction Fund and retirement savings fund Hostplus along with several other investors.

20.01 A bill cracking down on hate speech has passed the House of Representatives and Senate at a special sitting late on Tuesday, a month after 15 were killed at Bondi Beach. It includes provision to ban groups deemed to spread hate, the BBC reports. The hate speech reforms were originally part of an omnibus bill along with gun reforms, but the two were split last week. The gun reform bill, which includes a buyback scheme, was also passed.

16.01 Social media companies removed access to around 4.7 million accounts for those under 16 in the first half of December, eSafety says.

16.01 Alphabet’s Waymo robotaxi business has held discussions with Chinese car maker Geely, and other electric car makers, as it looks to enter the Sydney market. AFR reports. Waymo is stepping up its efforts in Australia, with the appointment of a lobbyist and a search for an office now underway. It has yet to lodge an application to begin testing in the city.

13.01 Advertising is hitting the AI chatbots with ads starting to show up in Google’s ‘AI Mode’ in what the Washington Post says is likely to be just the beginning of more trial-and-error attempts this year. Google has been pushing a new type of ad in AI Mode to advertisers, OpenAI has also been looking at introducing advertising in ChatGPT and Perplexity attempted ads but pulled back last year – but has left the doors open to try again.

08.01 The DTA has warned against outright bans of IT suppliers and services companies for unethical behaviour warning such bans could introduce significant operational risks and unintended consequences. ITnews reports parliament is considering an exclusion regime preventing those engaging in unethical conduct from bidding for government work, in the wake of the PwC scandal. The DTA, however, says blanket bans could block access to tech advances or skilled resources.

The post Appwrap: OpenAI infiltrates Medicare, Firmus eyes big $$$ and smart glasses ban appeared first on iStart keeping business informed on technology.

That challenge has been a recurring theme at Gartner’s IT Symposium/Xpo on the Gold Coast this week, with analysts saying that enterprises are becoming increasingly proficient at deploying AI, while remaining far less effective at managing its risks, costs and outcomes.

“Start by limiting careless consumption before it limits your critical use cases.”

In the opening keynote, Gartner analysts Kristen Moyer and Daryl Plummer painted a picture of organisations building momentum around AI without necessarily heading in the right direction.

“AI is the thing. Everybody wants it. Everybody’s gotta have it. Everybody’s building momentum with it. Or so they say,” Moyer told attendees. Yet while 60 percent of Australian and New Zealand executives say they’re getting positive value from AI (ahead of the 58 percent globally), only one in five AI projects generates positive ROI, a figure Plummer noted is unchanged from last year.

“The problem is AI risk is compounding much faster than value,” Moyer noted. AI use cases have increased more than 400 percent in a single year, but organisations are often ‘reinventing the wheel with every AI project’, rather than building on earlier successes.

At the same time, Gartner says executives are increasingly worried about the downside. Eighty-six percent of CIOs report risk growing faster than value, while only 27 percent proactively integrate technology risk and business risk.

Gartner’s answer is a concept it calls an ‘AI central bank’.

Drawing parallels with financial regulation, Plummer argued that enterprises need a central function that determines acceptable levels of risk, governs spending, coordinates decisions and establishes accountability across all AI initiatives, because AI is no longer just a tech problem.

The concept is a central governing body, with a ‘rotating chair’ between the CIO and CFO with permanent consultation from the CISO and others, that decides what AI is worth funding, what level of risk is acceptable and who is accountable for outcomes. Rather than allowing individual business units to pursue disconnected AI projects, the central bank would act as a control plane overseeing governance, economics, technology choices and shared intelligence across the enterprise.

“The AI central bank doesn’t regulate one flywheel in your organisation. It regulates them all,” Moyer said.

The concept emerges as Gartner sees a growing mismatch between AI deployment and AI governance, and the emergence of ‘an era of finger pointing’. Gartner figures show 65 percent of organisations expect to deploy agentic AI within the next 12 months, but only 27 percent currently have agentic AI governance capabilities in place.

“If you’re called to account and your answer is AI did it, boy, are you in trouble,” Plummer warned.

The pair pointed to examples including that of PocketOS, where an AI agent deleted the company’s entire production database and primary backups in just nine seconds in April, arguing that organisations’ need clearer ownership structures before AI becomes more deeply embedded in critical business functions.

“Regardless of who built the agent or who approved its use, when something goes wrong, who are they going to call?,” asked Moyer. “It isn’t Ghostbusters,” quipped Plummer.

Outcome, not consumption

The pair repeatedly returned to the idea that organisations are obsessed with AI consumption, rather than outcomes.

Moyer warned “AI doesn’t just create value, it also creates waste”, pointing to low-value AI-generated content and what Plummer called ‘careless consumption’.

“Like those AI generated dancing videos,” she noted. Every token, model call and GPU cycle consumed by frivolous workloads, comes at the expense of more valuable use cases.

“Start by limiting careless consumption before it limits your critical use cases,” she said.

The concern extends beyond infrastructure costs. Gartner cited research showing that 40 percent of workers received AI-generated work containing errors in a single month. Each occurrence took almost two hours to clean up, which the analysts calculated could equate to a US$9 million annual loss in a 10,000-person organisation.

“Start by limiting careless consumption before it limits your critical use cases,” Moyer said, arguing organisations need stronger control over AI spending and usage patterns.

“Vendors are desperate to sell and monetise AI,” Plummer noted. “They want you to build momentum for their products or just to buy more tokens. For you, that is bad momentum.”

He cited figures showing technology vendors are pitching productivity surges of 50 percent or more, while Gartner data shows just 16 percent in actual productivity gains.

The keynote also challenged assumptions about AI agents.

While vendors are encouraging organisations to deploy agents everywhere, Gartner warned that organisations are often solving simple automation problems with expensive AI tools.”AI agents are an inefficient way to do a title search in a relational database. Just use a function call,” Plummer told attendees.

A financial playbook 

In a later session, analyst Rober Naegle put the focus further on the economics.

Opening with a question about whether attendees would describe their AI budgets as ‘chaos’ or ‘control’ (a solitary hand for that one – ‘Quick, grab their business card’, responded Naegle), he argued that most enterprises are using cloud-era thinking to manage something fundamentally different.

“Cloud and AI are almost apples and oranges in comparison,” he said.

Cloud spending is largely driven by infrastructure consumption, capacity and utilisation. AI spending is shaped by prompts, interactions, outcomes, user behaviour and model selection. Organisations looking for the same tools, processes and funding models used for cloud, Naegle warned, are likely contributing to the chaos rather than reducing it.

One of the biggest mistakes, he suggested, is budgeting around tokens.

“If you’re trying to budget now for next year’s or two years’ utilisation, you’ve got to change the thinking,” he said, arguing that predicting token consumption and pricing years into the future is largely impractical.

Instead, he said, organisations should budget around value.

He proposed dividing AI expenditure into three primary categories: Personal productivity, functional priorities and business-critical capability.

Personal productivity includes individual use of copilots and assistants and currently accounts for roughly 60 percent of AI usage, according to Gartner’s observations. Functional priorities account for about 30 percent, while only around six percent of AI spending today is directly tied to business-critical outcomes such as revenue growth, cost reduction or risk mitigation.

Those figures reveal why many AI funding conversations continue to struggle.

For individual productivity tools, Naegle said organisations should focus on spend controls rather than extensive ROI analysis.

“I would tell you don’t waste your time,” he said of attempts to precisely calculate individual productivity gains. Instead, organisations should determine a spend appetite and manage within it.

Business-critical AI, by contrast, should be evaluated against measurable business outcomes.

“When it becomes business critical, I’m actually able to monetise and put that value on the balance sheet or the income statement,” he said.

The challenge is that many executives still struggle to articulate that value in financial terms.

Naegle cited Gartner research showing that when CFOs are asked about AI value, 74 percent point to time savings. Only five percent cite cost savings and six percent point to increased profits.

At the same time, CEOs continue pushing aggressive AI adoption agendas, often bypassing traditional business case processes.

“We’ve got to be AI ready. We want to be an AI-first company,” Naegle said, characterising the pressure many CIOs face. The result is organisations purchasing AI capabilities today without necessarily knowing how they will justify or fund them 18 months from now.

While Gartner’s keynote warned organisations to stop ‘compounding AI volume’ and start ‘compounding AI value’, Naegle’s session provided the financial playbook, arguing AI should be funded according to business value rather than model consumption.

The post Taming AI chaos with an ‘AI central bank’ appeared first on iStart keeping business informed on technology.

It’s not the kind of KPI most organisations put in their AI success stories, but for Rangi, HEB Construction head of digital transformation, the experience was a turning point. What began as an AI-powered resume project at the Kiwi civil infrastructure and engineering company became “a really good learning exercise” – and one that has resulted in a repeatable framework designed to minimise token consumption, measure value and provide a foundation for future AI projects.

“We used three quarters of an allocation of tokens for a quarter and now we’ve reduced it down to a couple each time we do the call.”

HEB, which is a subsidiary of France’s Vinci Construction Group, has multiple AI initiatives underway across contracts, HR, onboarding and infrastructure operations as it builds the governance, monitoring and data foundations needed for broader adoption.

The project that provided so many of those lessons began with a relatively mundane problem.

For decades, tender teams had manually assembled bios and resumes for bid submissions. Over more than 50 years in business, that meant multiple versions of employee records had accumulated across the organisation, making it difficult to easily identify the right people, qualifications and expertise for individual tenders.

The company implemented Boomi Enteprise Platform to build two AI agents. Shredder, deconstructs CVs and creates a structured representation of the person’s skills in an easily accessible and reusable format, while Bob is a conversational agent that retrieves appropriate profiles and generates a tailored CV or short bio for tender teams.

“It was a really good opportunity to clean up the process of how we manage our resumes and automate the data and have the ability to prompt and retrieve that data efficiently and with a high level of data integrity to be able to respond to tenders with the appropriate people that had those technical ability or that experience,” Rangi told iStart.

The project is part of a broader technology strategy built around Microsoft technologies including Fabric, Power BI, SharePoint and Teams, alongside Snowflake for data warehousing and Boomi for integration.

“We are wanting to use AI across any of the applications we have internally,” Rangi says.

HEB’s AI approach in New Zealand is centred on low-code and no-code development with support available from a wider pool of 88 data scientists at Vinci.

Everything starts with data

While the resume project delivered a practical use case, it also reinforced a lesson Rangi says applies to virtually every AI initiative.

“Everything starts with data.”

Construction is, by its nature, data intensive. Drone footage, compliance records, environmental monitoring, project information and health and safety reporting generate vast amounts of information. The challenge is less about collecting data than ensuring it is available in a form that can be trusted and used.

“Our ability to have our data in a place that we can actually use automation and AI tooling to automate and optimise processes is significant,” Rangi says. “It’s about optimising information to make timely decisions and freeing up our people to actually do the things that they need to do, rather than sitting in front of their devices trying to do reporting – which wouldn’t be their forte in a lot of cases.”

But while data is foundational, Rangi doesn’t believe organisations need to wait for a perfectly curated data estate before pursuing AI.

“You can’t clean up everything, it’s too big,” she says.

Instead, HEB has focused on identifying the information employees access most frequently and making that data available in a governed, usable format. As part of its migration to SharePoint, the company has been analysing how information is used across the business and prioritising the data most likely to deliver value.

“We’re working across the business to identify what data is accessed more frequently and enabling that to be turned into a data pipeline so our business can self-help and know the data is clean.

“For example, we use SAP for our ERP system and we’re using SAP Analytics Cloud and developing that to enable our business to safely access and do dashboards, but also we are starting to test Claude because the ability to call up dashboards on the fly instead of having to train people to use Power BI and get access to data quickly is something that we’re looking at as well.

“Technology moves so rapidly it’s not a one-stop shop and we might take an approach and then change that, but our our strategy is to enable our business to have access to data in a real, efficient and effective way and making sure the data is clean.”

The resume project highlighted the data challenge early. Before AI agents could retrieve information effectively, HEB first had to standardise how employee skills, qualifications and experience were represented. The company created a common template and loaded the information into Snowflake, creating a consistent foundation for the agents.

Token shock 

The bigger lesson emerged after deployment.

As users began experimenting with prompts, token consumption escalated rapidly.

“We soon experienced quite a lot of tokens being used for the prompting,” Rangi says.

The experience forced the team to revisit how information was structured and how requests were made. HEB refined the way data was organised and introduced more targeted prompting, reducing the amount of work the AI models needed to perform.

“We made the calls a lot smarter so that it could be more specific when they did their prompt plan,” she says.

The result was a significant reduction in token consumption – from that initial comment of using three-quarters of the allocation of tokens for a quarter, it’s now been reduced to “a couple” each time a call is made.

Perhaps more importantly, it has also resulted in a repeatable approach HEB believes can be applied to future AI projects.

“The way that we’ve done it is a repeatable way, so that framework or process that we’ve built, we can now look at reusing for other operational-type activity that we do quite regularly,” Rangi says.

Those opportunities are already emerging with “five or six” AI agents in testing or review and one to three AI-related initiatives worked on each month. Projects range from contract analysis and employee onboarding through to systems designed to help identify and forecast potholes on roads.

As AI activity has increased, so too has the need for governance and visibility.

HEB has established monitoring and alerting through Power BI dashboards to track token consumption across projects. The company is also investigating ways to provide visibility across AI use regardless of whether tools are built in Copilot, Claude, Boomi or other platforms.

“For each agent we build, we’ll have it in Power BI, and we can monitor those and have alerting,” she says.

“It’s given us the basis for having a very robust management solution across our agent use and visibility across our projects,” she says of the early learnings.

The company is also requiring business units to identify, with the tech team, expected returns before projects begin and agree on how those outcomes will be measured.

“A lot of AI initiatives don’t get out of test or development or get past the idea or concept because everyone wants to do AI so we’re doing a lot of validation upfront and ensuring that we identify how we can capture value.”

The value isn’t always measured in financial terms. Sometimes it’s time savings. In others, it may be a better experience for employees or easier access to information. The key, she says, is understanding what success looks like before development starts.

As HEB expands its use of AI across contracts, onboarding, asset management and other parts of the business, Rangi says the key lessons from the resume project have remained remarkably consistent.

Start with a clearly defined problem. Be realistic about your data. Measure outcomes from the outset. And don’t try to transform everything at once.

“Start with data, start small, support the journey by training,” she says.

The post HEB’s AI playbook and some reality checks appeared first on iStart keeping business informed on technology.

Netskope Threat Labs Australia and New Zealand 2026 report shows usage of the Model Context Protocol, a key technology underpinning agentic AI and enabling AI systems to directly connect with external applications, databases and enterprise tools, is rising rapidly across the region as organisations increasingly wire AI into operational workflows, software development environments and enterprise workflows, rather than using it solely as a standalone productivity tool.

“While MCP traffic is a sign that agentic AI adoption is progressing fast within organisations across A/NZ, each of these connections is a new pathway for company data to move between AI apps and outside systems…”

The findings align with trends identified in Netskope’s global 2026 AI report, which describes a worldwide shift from shadow AI concerns towards governance of autonomous AI agents and connected AI systems. Globally, Netskope says organisations are moving beyond monitoring prompts and are increasingly focused on how AI interacts with enterprise data and executes actions across connected systems.

Across Australia and New Zealand, one of the clearest indicators of that transition is the rapid growth of MCP, an emerging standard which allows AI systems to connect with external tools and data sources.

The report found the number of agents interacting with remote MCP servers increased by 89 percent during the two month reporting period, while MCP-related events rose by 69 percent. Globally MCP has also experienced a surge in use. The A/NZ report says much of this increased activity is linked to coding agents such as Claude Code and Codex, which are increasingly being used to interact with enterprise systems on behalf of users.

Rather than simply generating responses, AI systems connected through MCP can potentially retrieve information from multiple systems, interact with business applications and participate in business workflows. That represents a marked shift from the first wave of generative AI adoption, where employees primarily used AI as a productivity tool for drafting documents, summarising information and generating code.

The report warns that the same tools which unlock productivity, including MCP servers, simultaneously expand attack surfaces.

“While MCP traffic is a sign that agentic AI adoption is progressing fast within organisations across A/NZ, each of these connections is a new pathway for company data to move between AI apps and outside systems, and a vector of potential data leaks if not properly governed.”

The report says downstream data policy violations, where an AI service returns sensitive information to users or agents who are not authorised to access the information are rising as organisations connect AI systems to data stores and business applications. They accounted for 666 out of every 10,000 AI security alerts among organisations with the ability to monitor for such activity. Upstream data policy violations – where employees send sensitive data into AI tools, remained the most prevalent AI-related risk accounting for 8,300 of every 10,000 alerts across Australia and New Zealand.

Anthropic pulls ahead 

The growth of connected AI systems comes as AI adoption continues to surge throughout Australian and New Zealand organisations.

According to the report, the proportion of users actively using AI applications increased from 53 percent to 75 percent over the past year. However, direct AI use only tells part of the story, with Netskope finding that 97 percent of employees now use applications with embedded AI capabilities and 93 percent interact with systems that use customer or user data to train models.

“These figures illustrate how deeply AI has become embedded in everyday work, often through features built into familiar business applications rather than standalone tools.”

The report also points to a notable difference between A/NZ and broader global usage patterns: ChatGPT remains the leading AI application across most global regions monitored for the global report. Locally, however, organisations have shifted sharply towards Anthropic’s Claude platform which is now used by 81 percent of organisations in the region. ChatGPT follows on 68 percent, with Microsoft bringing up the rear with 66 percent. Netskope says a similar pattern is emerging in enterprise API usage, where Anthropic’s API leads local adoption, establishing a significant lead over competing providers. It has 82 percent of organisations connecting to it, well ahead of second-placed Assemblyai on just 47 percent and OpenAI on 44 percent.

That divergence may reflect the increasing influence of developers and enterprise users as AI adoption matures. The report notes that the popularity of Claude Code and related development tools is helping drive wider enterprise adoption of Anthropic’s platform.

Shortening shadows 

The findings also suggest ANZ organisations are bringing AI under more formal governance. Use of organisation-managed AI applications climbed from 34 percent to 75 percent during the year, while use of personal AI applications declined from 76 percent to 55 percent.

However, the report also highlights an emerging challenge with the proportion of users switching between personal and enterprise AI accounts almost doubling from 11 percent to 21 percent. According to Netskope, that indicates employees continue to experiment with new AI services even as organisations establish approved platforms and governance frameworks. Netskope’s take? It’s time to streamline the review and approval of new AI applications.

The post A/NZ companies wire AI into operations appeared first on iStart keeping business informed on technology.

As the capability moves into Auckland Council ownership as part of its council-controlled organisation reforms, Cohen, who is AT CTO, says the platform could be extended into new civic use cases and be monetised and sold to other agencies globally.

“It has provided significant value across different areas from enforcement to safety scenarios, and given us a bit of an operational edge to our business.”

Theia, which won the Cloud-native Development and Hybrid Cloud Infrastructure category at the recent Red Hat APAC Innovation Awards for Australia and New Zealand, has been under development for about a decade and today processes feeds from around 8,000 cameras across Auckland’s transport network, including road, rail and ferry. Created to help monitor roads, public transport infrastructure and other network assets, the platform is capable of automatic detection, tracking, data collection, identifying incidents and events and has evolved into a platform used for safety monitoring, enforcement, operational analytics and service management.

But Cohen believes its future could stretch well beyond buses, trains and traffic management. He told iStart he can see ‘millions’ of use cases for the technology in helping organisations identify problems earlier and respond before they escalate. One possible example: Identifying swimmers in trouble at public pools.

“With this moving to council, there will be even more opportunity and even more need because it will be across a much wider range of scenarios and use cases,” he says.

The CCO reforms will see the team responsible for the system moving under Auckland Council from October, leaving it up to the council what happens in future. Cohen though, is clear: “I personally think monetising the service would be a good idea as it would reduce rates for Aucklanders as a whole.”

Councils and public transport organisations around the world face similar operational challenges and need the capability of the system.

“It’s provided significant value across different areas from enforcement to safety scenarios, and given us a bit of an operational edge to our business,” he says. “From identifying traffic cones and where they might be sitting to identifying people who might be jumping on a rail network or scenarios where we’re currently working on aggressive behaviour detection to identify when aggressive behaviour occurs so we can send the right authorities.”

Building AI for Auckland 

Theia’s origins date back to a challenge many organisations continue to face today: How to monitor and keep track of all the assets and turn vast quantities of operational data into actionable insight.

Derek Zhang, Auckland Transport Computer Vision delivery manager, says the organisation initially relied heavily on manual monitoring of CCTV systems to oversee roads, public transport networks, parking facilities and other assets across the city. The approach was labour intensive, reactive and generated little usable data for decision-making.

After an early attempt to address the problem using a third-party platform didn’t work ‘for many different reasons’, Zhang says, and the decision was made to bring the capability in-house.

“We wanted to stand on the shoulders of giants like Google, Intel etc – I won’t mention too many names – and utilise what they have built, then modify and adjust it for what we need as a transport organisation.

The platform uses Red Hat OpenShift to provide the containerisation platform so ‘we don’t need to worry about the underlying technology that keeps the applications running’, Zhang says.

“It’s not just the AI component in it, but also how this system can be seamlessly integrated with our existing systems upstream and downstream.”

The project faced the unique challenge of operating in a real world and real time environment, requiring the technology to be reliable, scalable and available around the clock.

“That requires the technology to be able to cater for those requirements, how we design the infrastructure, how we design the product, how we link things together, without losing the requirement for real-time basically, and then it has to be maintainable as well,” Zhang says.

“We need to process those large volumes of data using the infrastructure and platform and then also need to maintain the trust, privacy, security and reliability, expected from a public sector organisation.”

Today, Zhang says the platform is used across multiple departments, including safety, compliance, network performance and asset protection teams. The system supports applications ranging from parking and bus lane enforcement to dynamic lane management, pedestrian analytics and rail safety monitoring. It is also being used in a pilot focused on detecting aggressive behaviour across the transport network.

The City Rail Link has further expanded the platform’s reach even further, with additional cameras and LiDAR systems being integrated into network operations. The technology is being used to monitor passenger numbers, assess platform occupancy  and identify people entering restricted rail areas or safety zones.

It’s also been used to enable ‘dynamic lanes’ – lanes which move between being bus only or open lanes depending on whether buses are running late or general traffic is building in order to keep transport flowing as smoothly as possible.

Going multimodal 

Cohen notes technology, whether AI, CCTV, machine learning or other mainstream offerings and even drones, underpins much of Auckland Transport’s KPI delivery.

The scale of the operation helps explain why AT is continuing to invest in the platform.

“We are talking about thousands of thousands of CCTVs that need to be monitored automatically by the system,” Zhang says, noting that reliability, scalability, privacy and real-time performance have become critical design considerations.

Those requirements are also driving the platform’s next phase of development.

While much of the current technology conversation has centred on generative AI, Zhang says Auckland Transport is preparing to incorporate both generative and multimodal AI capabilities into its roadmap.

Zhang says multimodal AI will help increase the accuracy level of the computer vision results while also improving interactions for transport operators and providing a better experience using the computer vision offering.

Theia complements a broader generative AI push underway within Auckland Transport. Cohen says the organisation is a significant user of Microsoft Copilot and is exploring ways large language models can be used to interrogate analytics systems and generate reports using natural language prompts.

 “We’re looking even at things like analytics that will be done through generative AI so that you can use an LLM to then generate a report to give you information you’re looking for, such as how many patrons did we have in the last 24 hours considering it was raining? How did that compare to last week when it was raining?”, Cohen says.

For Zhang and his 20-strong computer vision team, the bigger opportunity may be creating a reusable platform capable of serving an entire city. The environment that has been built is very scalable, and Zhang says expanding to new models would be ‘a lightweight piece of work because we have already developed the core product, which is going to run continuously as an end-to-end process.”

What began as an effort to automate CCTV monitoring is increasingly being viewed as a broader civic technology capability, one that could eventually support a growing range of council services while generating value beyond well Auckland’s transport network. And as Cohen says: “Anything that will reduce our rates, I encourage!”

The post AT’s AI ambitions go city-wide – and beyond appeared first on iStart keeping business informed on technology.

In a new report, Digital Foundations First: A Practical AI Pathway for New Zealand SMEs, the industry body warns that while AI adoption is growing rapidly, many smaller firms lack the connectivity, systems, skills and data foundations needed to translate AI use into meaningful productivity gains.

“The goal is not simply more technology spending, it is stronger business capability.”

The warning comes despite New Zealand ranking among the world’s most enthusiastic adopters of genAI. APEC’s (Asia Pacific Economic Cooperation) Business Advisory Council placed New Zealand third among APEC economies for genAI use among the working-age population at 39.5 percent in Q1 2026.

But TechNZ says widespread use does not equate to widespread business transformation.

“Most SMEs are still using AI in limited ways: Embedded tools and off-the-shelf applications for individual tasks, rather than integrating AI into core operations or redesigning workflows around it,” the report says.

SMEs risk being left behind

The findings mirror concerns in Australia where data from the National AI Centre shows 43 percent of Australian SMEs reported some level of AI adoption between December 2025 and February 2026, rising to 44 percent in February. However the Centre says more than half of SMEs had yet to meaningfully adopt AI and identified trust, perceived relevance and capability as key barriers to uptake.

The TechNZ report draws on discussions and research presented at an APEC workshop in Hanoi in July, where policymakers, business leaders and international organisations examined the challenges facing micro, small and medium enterprises in an AI-driven economy.

One of the key concerns emerging from the workshop was that AI capability is becoming concentrated among larger, better-resourced organisations. OECD research presented at the event showed AI adoption among small firms increased from around two percent in 2020 to 18 percent in 2025. However the gap between small and large companies remains wider for AI than for many other digital technologies.

TechNZ warns that if the trend continues unchecked, New Zealand could face a ‘two-speed economy’ in which larger companies accelerate productivity and competitiveness through technology while smaller businesses struggle to keep pace.

The report argues that policy setting should focus less on AI itself and more on the conditions that make successful adoption possible.

It points to a ‘4Cs’ framework developed by the World Bank that identifies four prerequisites for productive AI adoption: Connectivity, compute, context and competency.

The report says the priority is to help more companies build the basics that make adoption stick: Reliable connectivity, cloud and core business systems, quality data, cyber resilience, digital skills and the management capability required to change how work gets done.

“The goal is not simply more technology spending, it is stronger business capability,” the report notes. “A small business with a functional website, digital payments, cloud accounting, fit-for-purpose cyber security and usable customer or operational data is better placed to move beyond experimentation. It can use AI to create repeatable gains in sales, customer service, compliance, forecasting, administration and operations.”

TechNZ is calling for a staged approach to SME support, beginning with digital maturity assessments and foundational improvements in areas such as cloud systems, eCommerce, cyber security, data quality and workforce capability. Once those foundations are in place, businesses can focus on practical AI applications and deeper workflow integration.

Helping SMEs adopt low-risk, readily available AI tools that address a specific business problem – focusing on solving a problem, not adopting AI for its own sake is one of the steps highlighted in the report.

Adoption without transformation 

The report highlights significant capability gaps among businesses already experimenting with AI. OECD evidence cited in the report found 76 percent of AI-using SMEs were classified as novices, while 50 percent identified a lack of digital and AI skills as a barrier to further adoption.

“The message is clear: AI adoption is accelerating, but capability is not keeping pace,” the report says.

TechNZ argues the greatest opportunity lies not in creating more AI specialists, but in helping the broad base of New Zealand businesses develop the digital capability needed to become effective adopters. It’s urging a move from a one-size-fits-all model of support to instead providing a maturity based offer, ranking SMEs from novices to champions and tailoring support based on those, noting the largest public policy opportunity is at the front of the pipeline, helping novices build the digital capability and confidence to become effective AI adopters.

TechNZ is calling for New Zealand to prioritise a simple, visible offer delivered through trusted channels, pairing finance with advisory support, maturity assessment and implementation help, rather that grants alone, and ensuring learning is modular, practical and workplace-based.

Measurements should be based on sustained use, productivity, employment quality and market access – not attendance or pilots, and plain-language guidance on privacy, cyber security, data governance and responsible AI should be provided in order to create impact.

The report concludes that New Zealand should remain ambitious about AI, but disciplined about the pathway to productivity.

“The evidence from Hanoi points to a clear sequence: Build digital foundations first, support practical AI adoption next, and enable deeper transformation where firms are ready.”

The post TechNZ warns AI gap could create two-speed economy appeared first on iStart keeping business informed on technology.

In a new brief, FinOps for AI: From Managing Costs to Maximising Value, Bain & Co says AI is driving a new wave of technology investment that extends far beyond model and token costs, increasing pressure on already stretched tech budgets.

“AI layers a new, highly variable cost base onto technology budgets, often with limited visibility into where the money is going and what value it’s creating.”

It’s forecasting that by 2035, IT costs could increase by 75 percent in organisations that enable AI.  (That figure, however, is drawn from a ‘typical’ US$10 billion consumer packaged goods company.)

And it won’t be courtesy of poor cost controls either, with the consulting firm warning that even organisations which successfully manage tech spending should still expect costs to rise significantly as AI adoption increases, making disciplined investment decisions more important than ever.

“AI promises significant gains in productivity, innovation and growth,” the brief says. “But it also layers a new, highly variable cost base onto technology budgets that are already under pressure, often with limited visibility into where the money is going and what value it’s creating.”

The brief argues that focusing on token costs alone substantially understates the true cost of AI adoption, with AI increasing spend across multiple parts of the tech stack, including infrastructure, software, cybersecurity, data management and new operating models. Large language model workloads in particular are increasing demand for cloud, compute and software licensing.

“Token costs are only one part of the cost equation,” the brief notes, while acknowledging the growing numbers of companies burning through their annual token budgets by the middle of the year – “You’re not alone.”

Greater architectural complexity as AI models and agents are layered into already fragmented technology environments will increase integration efforts, governance requirements and operational dependencies and risk, while faster technology lifecycles will make investments obsolete faster, increasing the cost of keeping pace and cybersecurity investment is also increasing. Data and governance requirements will also increase adding to the investments required, while organisations are also being forced to absorb the costs of reskilling talent, redesigning processes and decision rights and operating legacy and AI-enabled models in parallel during the transition.

Cost, and cost cutter

For tech leaders, that’s creating a dilemma with pressure to deploy AI to improve productivity and drive business outcomes, while simultaneously controlling costs and maintaining governance.

But while Bain & Co describe AI as ‘the largest emerging source of technology cost’, they also say it is ‘the most powerful tool available to control those costs’ – enabling better visibility on spending through automatically classifying invoices, mapping ledger entries and identifying shadow IT; optimising infrastructure and helping rationalise applications; delivering software faster and creating more productive operations.

“Senior executives must deal with both realities at once,” the report notes. “Left unchecked, rising demands for compute, data, software, security and talent will erode margins; deployed deliberately, the same technology can simplify the estate, automate work and reset productivity.”

The challenge, Bain argues, is that many organisations still lack sufficient visibility into how AI spending translates into business value. That’s a view shared by Gartner, with a recent report showing 11 percent of organisations are ‘entirely unaware’ of what their function spent on AI in 2025 – but nonetheless are continuing to pump money into AI.

Bain says to meet the challenge, leading executives are extending FinOps beyond cloud cost management into a discipline for governing AI investment, linking consumption, technology costs and business outcomes in near real time.

Spend smart

The firm argues the goal should not necessarily be reducing tech spend.

“In this environment, the goal isn’t necessarily to reduce technology spending. It’s to direct capital investment toward the AI deployments that will create the greatest business value.”

Not every tech investment deserves protection, the report notes. “Where differentiation is real, protect investment. Where commodity delivery is sufficient, mandate efficiency.”

Technology costs need to be treated as a capital allocation decision, the company says – rather than an operational line item to be managed by IT. “It is one of the largest and most consequential investment decisions many companies make. Boards, CEOs and CFOs should oversee technology costs with the same rigour applied to capital allocation, requiring clear visibility into spending, AI investments, value realisation and emerging risks.”

Transparency across tech spend, including shadow IT, is the starting point, connecting costs to business outcomes. “Without transparency, cost reduction efforts often destroy value rather than create it.”

Bain & Co also recommends designing a ‘self-funding AI investment engine’ reinvesting AI-fuelled savings from the likes of application rationalisation and productivity gains, back into the highest-value AI opportunities.

It’s also urging companies to build a capability, not a program, saying tech cost reduction programs often deliver short-term gains only to see costs rebound within a few years. “Break this pattern by embedding real-time cost visibility tools, agile funding methods and ongoing accountability structures into the operating model. The goal is a permanent capability, not a one-time result.”

Winners ahead

The findings echo Bain’s earlier research into enterprise AI spending. In a separate survey of more than 400 technology leaders, Bain found 69 percent expected AI spending to increase by more than five percent as organisations accelerated adoption. The firm also warned that AI introduces additional complexity through faster technology cycles, more demanding architecture requirements and new operating models.

Taken together, the reports suggest that while AI may ultimately improve efficiency, organisations should not expect lower technology budgets in the near term. Instead, they are likely to face growing demands for investment as AI workloads scale and become operationally embedded.

For business leaders, Bain’s message is straightforward: The issue is no longer whether organisations will spend more on AI, but whether they can clearly identify which investments are generating business value.

“Companies that treat cost optimisation as a source of growth capital can reinvest savings into the AI capabilities that create competitive advantage,” the report says.

As AI spending moves from experimentation to enterprise scale, the winners may not be those spending the most, but those with the clearest understanding of what their AI investments are actually delivering.

The post The AI bill is coming: Why tech costs could soar 75% appeared first on iStart keeping business informed on technology.

The federal-government backed Buy Australia AI Partnership, launched last week by Stone & Chalk with the National AI Centre as principal sponsor, aims to connect Australian AI providers with major enterprise buyers, initially in the financial sector. Financial institutions ANZ, Commonwealth Bank, NAB, Westpac and payments provider Cuscal.

“It gives Australian AI companies direct exposure to enterprise expectations, while helping organisations better understand the capability available locally.”

The program is designed to help local providers build a deeper understanding of how organisations evaluate, procure, govern and deploy AI, in order to hlep them meet enterprise requirements, and connecting them with potential customers.

It’s a program designed to build Australia’s AI economy, or in the words of Lee Hickin, executive director of the National AI Centre, to ensure Australia is ‘the makers, not takers’.

“It gives Australian AI companies direct exposure to enterprise expectations, while helping organisations better understand the capability available locally,” Hickin says.

Organisations including the Australian Banking Association, Customer Owned Banking Association, Insurance Council of Australia, Australian Information Industry Association, Committee for Economic Development of Australia, MYOB and blockchain provider Solana have thrown their weight behind the program.

Andrew Charlton, Minister for Science, Technology and the Digital Economy, says there are already over 1500 Australian AI companies building ‘world-class technology’. “The challenge is getting in front of the organisations that can buy their products,” he says.

In a First Take on the initiative, Gartner says government-backed supplier ecosystems are emerging as one way to develop technology capability. Rather than relying only on regulation, subsidies or public procurement, governments can connect private-sector buyers with local technology providers and improve their ability to compete for enterprise opportunities.

Open-weight for enterprise

That trend is playing out as enterprises themselves reassess how they source and deploy AI.

A new Gartner report highlights growing enterprise interest in open-weight models, with the analyst company saying the performance gap between open-weight and proprietary large language models has narrowed dramatically and is now ‘close to disappearing altogether’. Driving that is the ‘runaway spending on AI and enterprise efforts to contain those costs’.

Unlike proprietary AI services accessed through vendor APIs, open-weight models allow organisations to download, inspect and modify a model’s core numerical parameters. Gartner says they are becoming an increasingly attractive option for IT-forward organisations and their token-hungry applications.

“Beyond potential cost savings, these models provide flexibility, control and independence from API-driven large language model hyperscalers and lock-in to their platforms,” Should You Ban or Embrace Open-Weight AI Models, by Gartner’s Darin Stewart, says.

“Open-source AI and weights offer substantial benefits over their closed, proprietary cousins,” the report says. “Open LLMs can lower switching costs, both financial and technical, reducing the risk of platform and vendor lock-in. Local or private cloud deployment, deeper customisation, resilience from provider dependence, and greater freedom to test, audit and optimise models for specialised work all indicate the unique appeal of open-source AI.”

The shift is also attracting attention in New Zealand.

Earlier this year, Amanda Williamson, director of Deloitte’s New Zealand Artificial Intelligence Institute told iStart Kiwi companies were beginning to recognise the strategic importance of the technology underpinning AI deployments. She highlighted open-weight models as one option for companies to consider.

“Until now, an approach that most organisations have been using is just switching on the model that’s provided in whatever technology stack they have access to, but there are other things that can be done, such as using open weight models,” Williamson said.

Greater control, greater responsibility

For enterprises, however, greater control comes with greater responsibility.

Gartner says organisations adopting open-weight models assume responsibility for ongoing management, including patching, monitoring, access controls, compliance and governance. The report also highlights concerns around data provenance, training bias, data handling practices and the removal of safety controls. “However, most commercial AI and LLM providers involve similar risks, just abstracted away from the customer and absorbed by the vendor,” Stewart notes.

The issue becomes more complex when foreign-developed AI models enter the mix.

Australia is among several countries which have placed formal restrictions on DeepSeek and other Chinese AI offerings. While Gartner says many of the concerns about Chinese technology are ‘overblown’, it acknowledges there are legitimate dangers than need to be taken into account and addressed.

It argues enterprises should evaluate models on their architecture, training data, capabilities and governance characteristics, rather than relying solely on country-of-origin considerations, saying organisations should adopt policies that are ‘nationality-aware, but nationality-neutral’.

For procurement leaders, the message on both fronts is similar: Expanding choice does not remove the need for due diligence.

While the Buy Australian AI program will enable proactive supplier discovery, and earlier visibility of emerging local providers, Gartner warns participation in Australia’s Buy Australian AI program should not be treated as proof of enterprise readiness, while the growing availability of open-weight models should not be viewed as a shortcut to lower-cost AI. In both cases, organisations still need to assess cybersecurity, governance, scalability and commercial risk.

But, it says, expect government-backed supplier development programs to become more common. It’s predicting that by 2028 at least five major technology markets will operate ecosystems linking domestic AI providers with large enterprise buyers.

The post Australian AI vendors get enterprise pathway, while open models rise appeared first on iStart keeping business informed on technology.

While explainability in AI has been a holy grail in recent years with regulators wanting it, governance frameworks demanding it and vendors increasingly highlighting it as a ‘key differentiator’ the research, which involved 228 experienced evaluators reviewing innovation proposals, found AI recommender tools were persuasive enough to convince the evaluators to reject decisions made by an expert human panel, leading them to pass up on promising innovations.

“A convincing explanation is not necessarily evidence that a recommendation is correct.”

Offer a narrative explanation – even for incorrect AI decisions – and the evaluators being tested were found to be even more inclined to defer to the AI decision.

The findings challenge one of the most widely accepted assumptions underpinning enterprise AI adoption: That providing users with greater visibility into an AI system’s reasoning automatically leads to better decisions.

Instead researchers found  that AI recommendations on their own improved decision quality, while adding a narrative explanation increased compliance without improving outcomes. Evaluators became more likely to follow the AI’s advice, but less likely to challenge it when it was wrong.

“We find that black-box recommendations [those provided without explanations] improve decision quality, whereas narrative explanations do not, despite inducing higher compliance.”

The study examined how evaluators responded to AI assistance while screening real submissions to a global social impact challenge run through MIT Solve, comparing human-only evaluation, black-box LLM recommendations and the same LLM recommendations paired with narrative explanations.

What stood out was not that people ignored the AI. In fact, the opposite occurred.

Both recommendation-only and recommendation-plus-explanation groups became more likely to follow AI guidance. But the addition of explanations pushed that deference further. Researchers found narrative explanations created an ‘asymmetric compliance’ effect, with evaluators becoming particularly likely to follow AI recommendations to reject ideas.

That mattered because some of those rejection recommendations were wrong.

The study found narrative explanations led evaluators to disproportionately follow rejection recommendations, ‘substantially increasing false negatives’. In the context of the experiment, false negatives were ideas rejected by evaluators that the independent expert panel believed should have advanced.

For organisations increasingly using AI to support decision-making, that finding may be more important than the research’s innovation-screening setting.

Significance beyond innovation screening

Across Australia and New Zealand, AI tools are moving beyond content generation. They are being used to evaluate funding applications, prioritise projects, screen job candidates, assess risk, review procurement responses and help staff navigate complex decisions. Many of those systems are specifically designed to provide explanations alongside recommendations, with explainability often viewed as a safeguard against blind reliance on AI.

The research suggests those explanations may sometimes have the opposite effect.

According to the paper, narrative explanations can suppress what researchers call ‘productive overrides’ – instances where humans correctly identify a flawed AI recommendation and choose not to follow it. Rather than encouraging greater scrutiny, the explanations may make the recommendation feel more authoritative and complete.

“Mechanism analyses show that narratives suppress productive overrides by substituting persuasive text for independent verification,” the paper states.

In other words, people stop checking.

The researchers argue that this is because large language model explanations function differently from traditional explainability tools. Instead of exposing the underlying logic of a decision, AI-generated narratives are designed to produce coherent and persuasive language. The explanations sound like reasoning, but may not actually represent the processes that produced the recommendation.

The paper notes that these narratives are often optimised for ‘linguistic fluency and persuasiveness’ and can create an ‘illusion of explanatory depth’ where users feel they understand a decision simply because they have been presented with a convincing rationale – the old issue of AI being very confidently wrong.

That has significant implications for current AI governance efforts.

Much of the discussion around responsible AI in recent years has centred on questions of transparency and explainability. Policymakers have pushed for greater visibility into AI decisions, while vendors have responded by adding increasingly sophisticated explanation capabilities to their products.

The study doesn’t argue against explainability and it doesn’t suggest organisations remove explanations from AI systems. Instead, it raises a more uncomfortable possibility: That explanations can increase trust without increasing accuracy.

The researchers conclude that effective human-AI collaboration depends on preserving independent human judgement rather than replacing it with persuasive machine-generated reasoning. As the paper notes, ‘LLM explanations do not necessarily improve decision-making’.

For business leaders rolling out AI across their organisations, the lesson may be straightforward. A convincing explanation is not necessarily evidence that a recommendation is correct. In some cases, it may simply make it harder for employees to disagree.

The post AI explanations may do more harm than good appeared first on iStart keeping business informed on technology.

Watson, Professor of Law at the University of Auckland Business School, says companies such as Amazon, Meta and Google no longer compete within markets. Instead they own essential digital infrastructure, control platforms on which people and businesses depend, and set the rules for operating within them.

“We have the opportunity perhaps to be bold within our own jurisdiction.”

She told iStart some platform operators hold a structural advantage because they both control the infrastructure on which markets operate and compete within those same markets, citing Amazon as an example.

“”They operate as the platform where people sell goods. And also, of course, Amazon itself is one of the players that sells goods on the platform,” Watson says. “That gives Amazon a big advantage because it both holds the infrastructure of the platform and is also one of the enterprises selling on it.”

She compared the situation to New Zealand’s ongoing debate over gentailers, where electricity companies both generate and sell power.

“It’s that same issue around how can you have a platform that’s fair for everyone when you have some players that are structurally advantages.”

Rather than relying solely on conventional competition measures, Watson says policymakers should consider structural remedies.

Changing the corporate

She says corporations are often misunderstood as essentially private businesses, when in fact they are legal entities, created and empowered by states. That misunderstanding has helped obscure the source of big tech’s power and contributed to the failure to rein it in.

“The sort of difference that would really shift what is happening is saying you need to look at the corporations themselves. One obvious way that you could reduce the power of these big platform corporations is that you could disaggregate where they are vertically integrated.”

“If they’re both the virtual infrastructure and a participant in the market, maybe we just say, ‘well, you can’t do that,” she says.

She goes further, suggesting platform operators may eventually need to choose between running the infrastructure and competing on it. “You’re going to be a platform or you’re going to participate on a platform – a corporation can no longer do both.”

But a broader way of thinking is to look at what we give to corporations when we enable people to incorporate them, she says.

Very large corporations such as the platforms ‘can create new types of capital and lay a claim on it in a way that regulators struggle to stop then doing’ she notes. “If you think about technology, they’ve done that with data – no one realised or claimed data because they perhaps didn’t see the value of it. But big tech corporations are essentially turning data into capital by recognising that data allows us to predict people’s behaviour, and that has value.”

She says that’s one example of what could be regulated against. Another option would be once platform corporations reach a certain size they have different requirements placed on them.

The proposal reflects broader arguments in Watson’s paper Reining in Big Tech Corporations: Why Platform Governance Requires Structural Regulation, which contends that platform corporations increasingly control essential digital infrastructure and should not be viewed simply as conventional market participants. She draws comparisons between the East India company and rail and oil companies of the Gilded Age in the late 19th century and today’s big tech. The paper will form part of a Cambridge University Press collection.

While many Gilded Age companies were broken up by antitrust laws, that involves the jurisdiction the company is based in taking action – something that she admits is unlikely to happen with large US-based companies.

Getting ahead of the problem

While critics argue stronger regulation risks stifling innovation, Watson says the challenge is developing targeted responses rather than broad-brush interventions.

“I always think it’s like a scalpel, not a sledgehammer.”

Regulation should focus on specific risks and structural characteristics, rather than applying restrictions across all large corporations.

Locally, she suggests Australia and New Zealand need to get better at predicting proactively, rather than reactively, the impact big tech behaviour might have, in order to consider whether precautionary approaches are needed before harms emerge.

“What always seems to happen with regulation is it happens after the harm has happened,” Watson says, citing the classing example of actions to prevent children using social media.

“We need to proactively think about, for example AI, and how might we predict what’s going to happen and then proactively prevent some sorts of activity within our jurisdictions.”

She acknowledges the answers won’t be easy.

“The only thing we could do is regulate activities of have platform corporations in our own jurisdictions that reach a certain size, or apply pressure on the international corporations as we are now with regulating social media, saying if you don’t do these things, we will in some way control access to our jurisdictions for your corporation.”

The discussion also has implications for New Zealand’s media sector.

Asked about suggestions social media advertising revenue sold into New Zealand could be levied with that funding used to support public-interest journalism, Watson says: “On the face of it, yes.”

The A/NZ opportunity

More broadly, she notes that major technology companies actively lobby against legislation they believe could affect their interests.

“It was interesting when they proposed the social media ban in New Zealand that Meta sent down a very senior official,” she says. “They will work actively to lobby against legislation that they think will harm their interests.”

Watson says New Zealand’s size should not automatically be viewed as a disadvantage. She points to the country’s history of leading on policy issues and argues smaller nations can still shape global debates – a factor that may have prompted Meta’s sending of that official. “Why do they care about little old New Zealand? Well it’s actually more likely that you’ll get that type of legislation in New Zealand than you will in the US or the big jurisdictions,” she says.

“We have the opportunity perhaps to be bold within our own jurisdiction and that might have some influence over other jurisdictions,” she says.

For business leaders, the central message is awareness.

Big Tech platforms deliver enormous benefits and are deeply embedded in modern commerce. But, Watson argues, their growing role as both infrastructure providers and market participants means businesses, policymakers and regulators need to think beyond traditional competition rules.

“If we could see them, they’re enormous forces,” she said.

The question, she argues, is no longer whether platform corporations are powerful. It is whether regulatory frameworks built for traditional corporations are still adequate when a handful of companies increasingly control the infrastructure of the digital economy.

The post Rethinking big tech’s power appeared first on iStart keeping business informed on technology.

That’s the contradiction at the heart of new Gartner research, which predicts AI inference costs per agentic workflow will increase more than fivefold by 2028 as organisations move from simple chatbots to more sophisticated AI agents capable of reasoning, planning and executing multistep tasks. The analyst company calls it the ‘inference paradox’ with improving model economics being overwhelmed by increasingly complex AI workloads.

While model prices might be falling, that’s luring users to build ever more complex workflows – and the greater token consumption of those workflows can outweigh savings from reducing model prices and escalating inference costs.

“The harsh economics of the Inference Paradox are exemplified by the differences between a simple chatbot and an AI agent,” Gartner senior director analyst Will Sommer says. “Where a simple chatbot must read and interpret a query and quickly respond with a probabilistically reasonable answer, an AI agent must constantly reason, negotiate and question itself.”

All of that processing comes at a cost.

Gartner says routing a task to an agentic reasoning model increases inference costs by at least five times compared with a basic chatbot interaction, and often much more as complexity grows. At the same time, organisations are discovering that increasingly capable AI systems consume vastly more tokens than traditional conversational AI.

Gartner argues many organisations are making matters worse by allowing inefficient agent behaviour to creep into systems as they scale.

“There is an enormous pool of waste in any agentic system,” Gartner says in a research note. “Most of the bill in agentic systems is redundancy.”

Among the issues identified in 50% of Token Costs Can Be Cut for AI Agents With Zero Quality Loss are repeated queries, redundant processing, oversized context windows and agent loops that repeatedly consume tokens without adding value. Gartner estimates roughly 31 percent of production queries repeat work organisations have already paid for. Agents consume around 100 input tokens for every output token generated, while 40 percent to 60 percent of agent tool output tokens can often be removed without any loss of quality.

This is where Gartner believes the next AI battleground is emerging as companies move beyond the first phase of experimentation and the second of governance, trust and responsible deployment and into AI engineering efficiency, or building systems that deliver the same outcomes using fewer resources.

That starts with model selection.

“Most tasks do not require frontier intelligence,” the report notes, warning that defaulting every workload to the largest and most capable models is ‘immensely wasteful’. Instead, Gartner recommends routing requests to the cheapest model capable of completing a particular task and escalating only when more sophisticated reasoning is genuinely required.

The report report estimates organisations can reduce costs by up to 60 percent through routing and workload optimisation alone.

Gartner argues that visibility becomes increasingly important as organisations scale AI into production. It recommends tracking metrics including cost per completed task, token consumption by model, spend by customer segment and the relationship between AI expenditure and customer outcomes. Without detailed measurement, organisations often fail to identify expensive models performing basic work or AI features generating little business value.

At the centre of Gartner’s recommendations is a concept that will sound familiar to cloud veterans: AI FinOps.

The company recommends introducing AI gateways, workload routing, caching, attribution and budget controls to actively manage AI consumption rather than simply paying whatever bill arrives at the end of the month. It describes routing, caching and orchestration as critical to preventing costs from spiralling as agent complexity increases.

“Hold agents to high-value tasks and delegate the rest down a tier,” the report says.

“Product leaders cannot rely on more efficient token economics to rationalise AI costs,” Sommer says. “Each successive generation of AI capability will necessitate more, and often more expensive, tokens.” He says there is ‘no reliable, economical one-size-fits-all model on the horizon’ and that organisations will increasingly need to manage complex multimodel environments.

The move by some AI providers to usage-based billing – as previously reported by iStart – has exacerbated issues.

Gartner’s report also identifies opportunities inside the architecture of agentic systems themselves.

Rather than feeding large volumes of information into models, the report recommends retrieval-based approaches that selectively surface only relevant content. Using retrieval-augmented generation (Rag), for example, can reduce token consumption by up to 75 percent while maintaining accuracy. Context compression approaches can reduce token volumes by 60 percent to 95 percent, while redesigning agent loops can remove 40 percent to 60 percent of redundant processing.

The broader message is that AI economics are changing.

For the past two years, the primary question has been whether AI could perform useful work. Gartner suggests the more important question now may be whether organisations can afford to run increasingly sophisticated AI systems at scale – and how to control what that intelligence costs.

The post AI’s new paradox: Cheaper models, bigger bills appeared first on iStart keeping business informed on technology.

At least that’s according to a new customer experience study which found Australian organisations are more than twice as likely as New Zealand organisations to be accelerating or expanding AI deployment in customer experience operations, at 71 percent versus 33 percent. Yet Kiwi organisations report stronger gains in productivity, revenue growth and cost efficiency from their AI-driven customer experience initiatives.

“Organisations have largely won the strategic argument around CX. The next challenge is execution.”

The findings come from the 2026 CX Capability Index, produced by Concentrix in collaboration with AWS and based on a survey of 545 senior decision-makers and customer experience professionals across Australia and New Zealand.

As the report bluntly puts it: “Buying technology isn’t buying capability.”

Of course, it should be noted that Concentrix is a customer experience outsourcer and services provider, and as such only to help to solve the issue of how to operationalise tech. Even so, the survey data it commissioned does support the broader argument that AI deployment alone isn’t creating differentiation.

Across Australia and New Zealand, 89 percent of organisations expect to increase customer experience investment over the next 12 months, up from 78 percent in 2025. Yet overall customer experience maturity has fallen from 70 to 66 over the same period and the proportion of organisations classified as CX Leaders has dropped from 28 percent to 21 percent.

The result is a growing middle ground. Sixty percent of organisations now sit in the CX Follower category, up from 46 percent a year ago, while one in four organisations that previously qualified as CX Leaders have now slipped backwards.

Everybody’s spending, but fewer organisations are pulling ahead.

Australia remains ahead of New Zealand on most customer experience metrics. Australian organisations scored 74 on the overall CX Capability Index, compared with 67 for New Zealand. Australia also reports higher customer experience maturity, stronger current investment levels and a greater proportion of CX Leaders.

But despite New Zealand’s company’s slower acceleration of AI in customer adoptions, those who already have live CX initiatives are outperforming Australia on productivity gains (61 percent vs 54 percent), direct revenue growth (57 percent vs 46 percent) and cost efficiency (44 percent vs 39 percent).

The report’s most interesting finding may be what happens after organisations deploy AI, and Concentrix isn’t alone in noting that adoption is no longer the main challenge.

Recent McKinsey research found many organisations are investing heavily in AI, rolling out tools across the workforce and encouraging widespread experimentation, but still failing to create enterprise-wide value. McKinsey highlights that AI does not create meaningful business outcomes simply because more people use it. Organisations generating the strongest returns are redesigning workflows, operating models and ways of working around the technology, rather than treating AI as another software deployment.

Boston Consulting Group also reached a similar conclusion in earlier AI Radar research which found three-quarters of executives rank AI as a top-three strategic priority but only one-quarter report generating significant value from AI investments. It found the companies achieving the greatest returns focus on a small number of high-value initiatives, scale them quickly and redesign business processes around them.

The same pattern is showing up closer to home: Datacom’s 2025 State of AI Index found 87 percent of Kiwi organisations are now using AI in some form and 88 percent report positive operational impacts. But only 12 percent had successfully scaled AI across their organisation. Nearly half remained in exploratory states despite widespread adoption.

That challenge of operationalising AI and gaining true value is visible in the CX Capability Index. While 63 percent of organisations are accelerating or expanding AI deployment in customer experience, 93 percent report barriers to deploying or scaling it effectively.

Cost leads the list at 45 percent, followed by privacy, compliance and trust concerns at 44 percent. Skills shortages and data quality issues each rate 38 percent.

The report argues the biggest obstacles are organisational, rather than technical with governance, operating models, workforce capability, internal resistance and strategy emerging as larger barriers than the technology itself.

That finding is reinforce by what the report describes as a widening execution gap.

Strategic commitment to customer experience increased 75 to 77 over the past year. Investment intentions strengthened. Yet the gap between strategic importance and maturity more than doubled, increasing from 4.4 points to 10.7 points.

In short, organisations understand the opportunity. Delivering on it is proving harder.

The report also challenges a common assumption about how AI will reshape customer-facing operations. The highest ranked AI use case was not autonomous customer service. Instead, 34 percent of respondents identified real-time support for frontline employees as the area where AI creates the greatest value. Hyper-personalisation ranked second at 20 percent, while autonomous handling of routine customer enquiries ranked lower.

“The strongest use cases are centred on employee enablement rather than customer automation. This suggests organisations currently see AI less as a replacement for frontline teams and more as a tool that enables employees to deliver stronger customer experiences,” the report says.

That ties in with 2025 results which saw a shortage of talent cited as the biggest barrier to improving customer experience. “Organisations appear to be directing AI toward last year’s most pressing constraint – augmenting scarce frontline capability, rather than replacing it.”

While the report issues a warning for New Zealand, saying the country’s slower rate of adoption risks seeing it fall ‘structurally behind, not just cyclically’, there’s also a warning for Australian organisations: Leading New Zealand on AI deployment hasn’t automatically translated into stronger business outcomes from AI-powered customer experience initiatives.

“The findings suggest organisations have largely won the strategic argument around CX. The next challenge is execution,” the report says. “Those that successfully translate strategic commitment into sustained organisational capability by treating CX transformation as an operating model rather than a technology procurement decision will be best positioned to differentiate themselves.”

The post Australia wins AI adoption, NZ wins results appeared first on iStart keeping business informed on technology.

A new discussion paper from the Helen Clark Foundation and New Zealand Rural Land Company argues that information collected from sensors, drones, satellites and connected farm equipment is rapidly becoming a commercial asset, creating tension between farmers who generate the data and tech companies that use it to build products and services.

“Farmers may be sceptical of third-party access to data they have generated. Innovators, meanwhile, are incentivised to build proprietary datasets.

The report, Growing Innovation: The Next Wave of AgriTech for Rural New Zealand, identifies agricultural data as one of five critical themes to ensure that the benefits of agritech revitalise rural communities and deliver maximum economic and sustainability outcomes for New Zealand. It warns a lack of consensus around ownership, access and commercialisation could become a barrier to innovation and adoption.

“Data is the crown jewel of 21st-century commerce,” the report notes, highlighting the value being created as farms become increasingly digitised.

The report, based on interviews with farmers, co-operatives, levy bodies, venture capital firms and public sector leaders, highlights agritech not just as a productivity and efficiency booster, but as an environmental sustainability enhancer. A World Economic Forum study found a 20 percent increase in agritech adoption in the EU could improve farmers’ livelihoods by €1.9–9.3 billion annually, improve soil health by 14 percent and reduce emissions by six percent by 2030.

Today, data is being captured across New Zealand farms through technologies ranging from environmental sensors and livestock monitoring systems to drones and satellite imagery. Agritech companies use that information to develop tools that improve irrigation, fertiliser use, emissions reduction, animal welfare, product quality and farm productivity.

But as the commercial value of that information grows, so too do questions over who should control it and who should profit from it.

The report says farmers and technology providers are becoming increasingly intertwined in the data economy. Farmers generate the information, while agritech companies use it to train algorithms, refine products and create services that can be commercialised across the sector.

The relationship is not always straightforward.

“Farmers may be sceptical of third-party access to data they have generated, concerned about intent and ownership,” the report notes. “Innovators, meanwhile, are incentivised to build proprietary datasets to gain a competitive advantage by locking up data, or to generate new revenue streams by charging a fee for access.”

That can leave farmers unable to access insights derived from collective datasets, despite contributing to the underlying information.

That dynamic can create barriers to agritech adoption, the report notes.

Aiden Gent, ASB general manager rural banking and a contributor interviewed for the discussion paper, says the industry remains divided on fundamental questions around data control.

“There are people competing to own data and commercialising on the fact that they own that data, rather than an open and free market,” he says.

The debate mirrors broader discussions unfolding across industry as organisations wrestle with how AI systems, analytics platforms and software vendors generate value from data supplied by their customers.

According to the report, there is currently no agreement within New Zealand’s agritech ecosystem on who should own agricultural data.

Some stakeholders favour a data sovereignty model, arguing that farmers should directly benefit because they generate the information. Others see data as a commodity that belongs to those who collect, process or commercialise it.

“These perspectives highlight the need for clear frameworks that balance commercial opportunity with fairness and trust and maximise value to the New Zealand food and fibre sector.”

The discussion paper argues that resolving those questions will be critical if the sector is to maximise both innovation and trust.

Nick Rowe, head of customer innovation at Silver Fern Farms, says the challenge is creating mechanisms that reward both parties.

“The challenge is creating a two-way street by connecting supply of high-integrity data generated by landowners, with market demand for verified farm-level data products and enabling an exchange of that data which creates value for both sides.

To address the issue, the report recommends establishing an Open Data Insights and Benchmarking Platform that would aggregate and manage agricultural data on behalf of farmers.

Under the proposal, farmers would be able to benchmark performance metrics such as yields, emissions intensity and water-use efficiency against regional and national averages without exposing sensitive information. Researchers would gain access to aggregated datasets, while agritech companies could use the information to support product development and innovation.

The platform would operate as an independent repository governed by agreed principles around data sovereignty, transparency and de-identification. Rather than individual farmers negotiating separate arrangements with multiple technology providers, the platform would manage licensing and access arrangements on their behalf.

The report also proposes that revenue generated through commercial licensing of aggregated datasets could be returned to participating farmers and reinvested into the platform itself.

The recommendations come as New Zealand’s agritech sector continues to expand. The paper cites MBIE estimates suggesting there are more than 500 agritech companies operating in New Zealand, generating annual revenue estimated at between $2 billion and $3 billion. Technologies spanning automation, AI, biotechnology and precision agriculture are increasingly being deployed across the food and fibre sector.

The report argues that agricultural data will sit at the centre of that growth.

The bigger agritech push

The issue of data ownership is one of five themes identified in the report, which argues New Zealand risks leaving economic value on the table unless it accelerates agritech adoption and addresses barriers to innovation.

It says the sector’s potential is bing held back by financing barriers that favour large operators over family farms, gaps in rural connectivity and services, the fragmented data ownership and a slow regulatory environment.

Among the themes identified is designing homegrown technologies which also target global challenges to capture international markets and creating innovation-enabling conditions with connectivity, risk-sharing, adaptive regulation and targeted incentives to accelerate adoption of new technologies and de-risk investment.

The paper outlines 13 recommendations aimed at strengthening the sector, including modernising agricultural education with digital and AI skills, creating new agritech micro-credentials, establishing a digital extension service to improve technology uptake and improving rural connectivity.

It also proposes a co-funded Agritech Adoption Fund to help farmers share the risk of investing in new technologies and a regulatory sandbox to enable faster testing of emerging innovations.

The report argues that New Zealand’s food and fibre industries are generating increasing amounts of valuable data while facing growing pressure to improve productivity, sustainability, traceability and compliance. In that environment, technologies such as AI, automation, precision agriculture and digital farm management are expected to play a larger role across the sector.

Among its more ambitious proposals are a national “Living Knowledge Bank” to capture farming expertise and an open benchmarking platform that would make agricultural insights more widely available while protecting individual farm data.

The paper’s broader message is that technology alone will not deliver better outcomes. Investment, skills, trust, connectivity and data governance will all be needed if New Zealand wants to convert agritech innovation into productivity gains and export growth.

The post Agritech’s data battleground appeared first on iStart keeping business informed on technology.