Insurance News magazine August/September 2026

Change agents

Our tech conference hears how self-directed AI could transform operations

By Miranda Maxwell

“The future is not humans versus technology – it is humans using technology better to strengthen the human side of insurance, not replace it.”

That was the conclusion of Pinnacle Connect founder Montsie Valdez, one of almost 200 professionals who gathered at the Insurance News Beyond the Buzzwords conference in July to hear how breakthroughs in technology are reshaping the industry.

This year’s sold-out event featured sessions dissecting the “agentic-to-agentic” future of self-directed AI, the rise of deepfakes and a 2030 insurtech road map, plus masterclasses on electric vehicle adoption, aggregation risk and quantum computing.

There was also a lively debate about whether technology can bridge the trust deficit in insurance. Below are some thought-provoking highlights from the event:

Allianz chief information officer Joanne Walker

In banking, there’s been a bit of a migration back to data centres – a lot of conversation around maybe keeping your data assets on-premise so you can really have that protected. Insurance is still moving to the cloud and things might actually end up coming back at some point.

Everyone thinks we’re going to move into this lovely, cheaper world, and hopefully that might be the case. But there’s also a fairly significant financial risk once we federate out, we remove our assets into the broader ecosystem. And the cost of the large language model will be kind of using a sledgehammer at the moment. I think there’s going to be more focus on: is the underlying AI model or set-up you’re using commensurate with the process you need to do?

BizCover CIO Dino Tius

Moving forward, the problem the industry has to solve is: how do we do agentic to agentic? That’s a real question.

This is pretty scary, to think about what will follow in the future where the agentic does the pricing and the underwriting, and also improves over time. So if it makes a mistake, it knows. They’re making some progress in the US in financial services. There’s a pilot to purchase … we reached out to OpenAI and they said, “We think we’re at least 12 months out before there’s a solution in Australia to continue the journey and purchase [insurance policies] with ChatGPT.

Regulation is definitely a challenge – which licence does it buy? – but I think also a lot of it is trying to figure out how they’re going to monetise this.

QBE Ventures head of emerging technology Alex Taylor

The industry will start to succeed using something that, frankly, is a little bit scary … but you need to understand precisely how the system can make a decision, and whether it’s making the right decision.

As much as we’re handed these amazing tools, suddenly we need to dramatically increase our awareness and understanding of our risk of operating. How do we know it consistently outperforms a human for this task? When we can do that, the sky’s the limit for what we can achieve with this [agentic AI] technology. But we’re not there yet, we have to be careful … coming up with methodologies to adapt to being able to acquiesce to change, but at the same time looking at the downside risk.

Endava senior insurance lead for Asia-Pacific Baneesha Narang

The next competitive divide in insurance will not be access to AI, but the ability to operationalise and scale. Those that lead will apply AI deliberately, creating measurable value while preserving human judgment.

Endava senior vice president for AI and technology advisory Mathew Finch

The insurance industry doesn’t need more excitement about what AI might do.

It needs a clearer path from a promising use case to something that actually delivers value, with the right data, governance and accountability around it.

WTW cyber and technology industry leader Benjamin Di Marco

When you load up your environment on your terminal at work, there’ll be hundreds of APIs running within your organisation between third parties. They’re described sometimes as the webbing of modern architecture, essentially how we stitch things together.

Why are insurers worried about it? What we’re seeing more and more is there is such a concentration of technology providers that any type of outage can impact all sorts of organisations across the supply chain, and that’s the theme … with aggregation exposure.

In the Optus incident that happened a few years ago, that API should have been decommissioned long beforehand. The fact it wasn’t was a huge part of the problem.

Really, if you’re thinking about API concentration, it is difficult to decouple … if some fail, the system can still work. That’s the thing that’s driving lawmakers, insurers and risk. APIs are designed to be one to many … that’s why it’s so difficult to manage.

All of these AI systems we talk about, they’re going to be using APIs … and they need to have broad access. I think the agentic stuff will move to MCP – model context protocol – but I think traditional environments will have APIs for a long time because you fundamentally need the connectors for modern architecture to work.

So these connection problems that we’re talking about, they’re going to essentially continue to repeat. They’re going to be a huge challenge for us to address.

Actuaries Institute council member Anthony Lowe

Portfolio optimisation, risk assessment and Monte Carlo simulations will all, in time, benefit from quantum computing. There are opportunities for insurance for sure, but we need to take a bit more immediate action on post-quantum cybersecurity.

The transition to post-quantum cryptography is going to take time – the keys that are used are much, much larger and the new standards do require more processing power. So it does take some planning.

There is credible evidence that mappers are already capturing encrypted data, and it’s not just so-called “harvest now, decrypt later” attacks, but also digital signatures. So “trust now, forge later” attacks as well, where you’ve digitally signed a legal document and then some time later somebody comes along and orders the document and forges it.

IVAA managing director Matthew McHutchison

We specialise in jewellery claims, hearing aids, glasses and furniture, and we are seeing fraud at an all-time high. Right now, with our insurer clients, we’re preventing more than $2 million of fraud every year … It is a growing concern we’re seeing on a daily basis.

Truuth co-founder Mike Simpson

AI-altered images are growing at scale. These trends didn’t exist a couple of years ago, so it’s a new challenge for us as an industry, and it requires some new tools.

Do an audit. Load 1% – or 100% – of your documents over the last year and check how many of them are repeat or close images, how many are deepfakes or have anomalous metadata associated. That’s beneficial because most insurers don’t know the size of the problem, and it’s a very simple check to run on how big the fraud problem is and how much it’s growing month to month.

Instanda general manager Asia-Pacific Jon Moody

Every scheme arrives with its own product rules, distribution channels, pricing logic and servicing workflows. The traditional cost of moving smaller schemes onto a centralised platform hasn’t justified the benefit.

We actually make things really efficient when we can get products that may not ordinarily be economical come onto the platform. We adapt our approach and style depending on what we find.

We, of course, will utilise tools like AI to reverse-engineer data collection points from PDSs and all that sort of great stuff, which is speeding up the process that used to maybe take a few more workshops and conversations. So we compress the time.

The Instanda platform has one set of rails, continuously configured, new schemes and changes without external delivery queues, with new features every six weeks.

Entsia founder and chief executive Alistair McElligott

What drew me to the insurance industry was just the complexity – the joke was, banking’s complicated. Insurance is more complicated, but insurance is 10 years behind the banks in terms of technology.

You’ve got to embrace the necessary complexity in insurance … but I think we add unnecessary complexity in a lot of areas. One of my daughters was turning 16 … I called our company broker … and she came back and said, “I’ve gone through everyone, and there’s only one [motor insurance] company that will cover you.”

It was actually a very complicated process … even for someone who’s in the industry and understands it quite well from the inside. It was an interesting experience having a hard-to-place risk, like a 16-year-old daughter.

Insurance Advisernet systems general manager Steve Dymond

People out there actually want to be able to pick up a phone and ring somebody they know when their warehouse is burning down. That is the most valuable thing we hold as brokers – real relationships and real people are very, very important. If we get AI to just get all the information, the bit that’s missing is the trust. The million-dollar question, really, is how do you find that balance?

BTB was organised with Insurtech Australia and backed by gold partners Endava, Instanda and Entsia. See more pictures from the event here.