Guidewire: Modernising insurance with accountability

AI might be a set of new and exciting tools, but some things don't change. Into that category Guidewire's Senior Director of Product Management for APAC James Couzens firmly placed accountability, compliance, and a laser-focus on the human factor. He said any product development always starts with the company's customers, and advances within the broader themes of industry norms and regulations. "I don't know if you've seen, but recently APRA put out a notice for insurers, basically saying 'hey, we know all this AI stuff is going on, but we still hold you accountable for what your AI does'."

While he said it with a smile, the underlying message is crystal clear: APRA, the body tasked with ensuring a stable, competitive and efficient financial system in Australia, is vigilant, and the same vigilance can be expected from authorities and regulators in every jurisdiction. 

As is by now equally clear, while AI might do things fast, it doesn't always do them right. Recent history is littered with examples of unfortunate or careless professionals taking AI shortcuts to their detriment, and it is always the person who blushes (or even sees their career end in tatters), and not the AI.

Guidewire is a Silicon Valley-founded company now operating in 43 countries, providing the software platform insurers need to run their businesses. As Couzens related and repeatedly stressed, Guidewire focuses on delivering what its customers want: claims processing was the pain du jour two decades ago, so that's where it got its start. The company has since expanded into policy administration, billing, data and analytics, and cloud services, and now, unsurprisingly perhaps, it is helping drive compliant AI adoption.

Couzens provides an illuminating peek into Guidewire's product development process. "We don't build new capabilities without customer validation to make sure that we are building the right thing," he said.

It doesn't happen in isolation, because developing one-off features or functionality is costly, both immediately and owing to ongoing support implications, long-term. "Fortunately we've got a great group of customers who are very willing to talk to us. We may have one customer we're working directly with, but we also have a number of others which we will be able to go back to and say okay this is what we're doing going to be of value to you? If there was something you want to change, what would it be? As a result, we actually make sure we have customers going on the journey with us."

There is, he added, "Nothing worse than building something that's not going to be used. So, we want to make sure, if we're going to build something, we know it's going to be valuable to the market."

These observations from Couzens are worth noting, particularly in the AI gold rush. Any hyped technology comes with early adopter risk (again, the missteps are well-publicised), and the trick is gaining the benefits while avoiding the pitfalls.

He said the risk aversion central to the premise and practice of insurance stands the industry in good stead. "One of the key things is that insurance is focused on risk, so while they see the value that AI will introduce, it's very much underpinned by accountability and an appreciation that there has to be a human in the loop."

That human isn't just the underwriter, either. It is also the regulators and, ultimately, customers. "You can't have AI making decisions without anyone being responsible and accountable. And you have to do things in a way that your regulator is not going to frown upon," Couzens emphasised. "APRA has definitely said yes, you need to be able to back those [decisions] up otherwise you'll be in trouble." Recently released products like Guidewire's ProNavigator, now embedded in InsuranceSuite, are uniquely placed to leverage an insurer's own knowledge and surface this in an auditable way when staff need it most.

He said AI is expected to deliver the next wave of operational efficiency, with particular opportunities in high-volume 'busy work' areas, with one example being document ingestion for workers' compensation claims. He said AI is eminently suitable for extracting data from medical certificates, reducing manual work and allowing staff to focus on injury management and recovery. In this administration-heavy business, there are countless similar use cases.

With the insurance industry grappling with rising costs, contributing to a problem of affordability for under-pressure customers, Couzens said AI powered efficiency gains can meaningfully lower overall inputs, potentially easing premium pressure.

But wherever AI takes the industry, Couzens said foundational constants remain. "We take accountability seriously," he said. "It's built into our corporate DNA."