AI in August

Explore how organisations across Aotearoa New Zealand are applying AI in real-world settings, alongside emerging international developments in AI transparency, autonomous agents and robotics. Discover upcoming learning opportunities and insights from members helping shape the responsible use of AI in engineering practice.

Upcoming AI webinars and workshops

Engineering New Zealand webinars and workshops

Join the Electrical Engineering Group online for AI as a productivity tool in an electricity utility: A personal journey. Dan Martin, Chartered Professional Engineer at Essential Energy, will share practical examples of using AI to support everyday engineering tasks, including analysis, coding, reporting and asset management. He’ll also discuss the governance, risks and professional responsibilities that engineers need to consider when using AI in practice.

Our next full day in person workshop, Exploring the Power of Generative AI, is taking place on Tuesday 15 September in Christchurch, kickstart your engineering and AI journey, register now.

Missed our latest Panel Q&A webinar? Watch the discussion on how universities are responding to the opportunities and challenges AI brings for engineering education.

AI Forum New Zealand update

The AI Forum’s AEC Working Group continues to produce practical material for people working in engineering and the built environment Knowledge Hub Archive - AEC Working Group.

The Aotearoa AI Summit is being held in Wellington on 8 and 9 September. Registrations are still open.

Engineering and Aotearoa related AI developments

Digital twins transform telecommunications infrastructure. FortySouth has won the 2026 New Zealand CIO Innovation Award for emerging technology for its digital twin programme covering its network of more than 1,700 telecommunications towers. The system combines drone imagery, detailed 3D modelling and AI-enabled analysis to create virtual representations of physical assets. A process that could previously take months can now provide an initial assessment in around 10 minutes, including checking available tower space and structural capacity. Formal engineering calculations are still undertaken before work proceeds.

Christchurch company applies computer vision to construction take-offs. Christchurch-based TakeoffQS is using computer vision and AI to automate parts of residential construction quantity take-offs. The software identifies and measures elements in building plans, converts verified measurements into material quantities and supports pricing and quoting. The company says its AI has been trained specifically on New Zealand residential plans and importantly, users review and verify measurements before they are used. The product is now being used commercially, with other New Zealand construction businesses running pilots.

AI listening devices could support Predator Free 2050. AUT researchers are using AI to improve detection of the last remaining possums in areas undergoing predator control. Small battery-powered recording devices can run AI models locally to recognise possum calls, avoiding the need to upload large quantities of audio. The researchers developed a new training approach that substantially reduced false detections when the models were tested against previously unseen New Zealand forest recordings. Beyond conservation, the work is a good example of edge AI, where models operate directly on sensors and devices in the field.

AI trends to watch

AI transparency rules kick in across the EU and California. On 2 August, new AI transparency requirements came into force in both the European Union and California. The EU AI Act now requires greater transparency around interactions with AI and AI-generated or manipulated content, while California’s AI Transparency Act requires large generative AI providers to provide detection tools and embed provenance information in generated images, video and audio. The alignment reflects a wider shift towards making AI-generated content identifiable and traceable, and is worth watching as these requirements increasingly influence the tools and services available globally.

Interesting reading

The Rise and Fall of Agent Civilizations. This article takes a closer look at the recent Hugging Face breach and the wider OpenAI agent incident behind it. During internal evaluations, groups of AI agents found ways to communicate and coordinate, with hundreds ultimately participating in attacks on Hugging Face infrastructure. The article also looks at what happened next, when a later group of agents gained administrator access to part of OpenAI’s own research infrastructure. It raises important questions about testing, monitoring and safeguards as AI agents become increasingly autonomous.

Robotics on Richard's radar

This month, I have been fascinated by the Chinese Humanoid Robot Olympics – both the successes and the failures. However, there is more to this than memes and TikTok videos. This piece looks at the bigger issues behind the Olympics.

I also continue to follow Boston Dynamics and enjoyed the technical answer to the question, why does Atlas have a head?

And finally, though not directly robot related, I note LinkedIn has now given users the ability to mark posts as “Seems like AI slop”. A good step for the future, I think.

Member spotlight: Dr Amir Mohammadi

“You can't sign a PS1 on a probability. AI is only as good as the engineering foundation beneath it.”

Amir joined the AI Advisory Committee earlier this year and is Chief Technology Officer and co-founder of Nodey, where he is helping rethink how residential structural engineering is delivered. Nodey turns architectural drawings into structural designs, calculations and consent documentation, compressing work that has traditionally taken engineers days to complete into hours. He and his co-founder started it after years of watching capable engineers spend most of their week re-typing numbers between spreadsheets on work that was well understood and endlessly repeated.

At Nodey, Amir is taking a deliberately different approach to AI in engineering. The platform is built as a deterministic calculation engine first, and every report includes the equations, design values and clause references so a peer reviewer or BCA can verify the calculations with full transparency. AI then sits on top of that foundation, helping with the judgement-heavy work around the design rather than the calculation itself. His key lesson from this approach: the sequence is everything. Build the verifiable engine first and AI becomes a real multiplier on engineering practice.

Let us know what you think, what you’d like to see more of, or share a success story of your own – get in touch at ai@engineeringnz.org