[00:00:00] Cameron Scott: Trust is what we see as taking AI deployment to the next level and being something that really helps an insurer run their business.

[00:00:08] Paul Lucas: Hello, everyone. Welcome to Insurance Business TV. And we’re here to talk about the industry’s hottest topic, AI. Of course, as AI adoption in insurance matures, the conversation is shifting from speed to trust. The real measure of value is no longer how quickly information can be processed, but whether the output can be relied on within core workflows such as underwriting. In insurance, that requires a higher standard, and one is helping to deliver that. That man is Cameron Scott, VP of Sales and Marketing at Insurance Systems Inc. Cameron, welcome to Insurance Business TV.

[00:00:41] Cameron Scott: Thanks, Paul. Great to be here.

[00:00:42] Paul Lucas: So Cameron, many insurers have focused on AI for speed and efficiency. Why is trust now a critical benchmark? And indeed, what results are your customers seeing with ISI AI in production?

[00:00:55] Cameron Scott: Speed and efficiency really got the industry to experiment and experiment quickly with AI, using Copilot or ChatGPT on another monitor to aid in day-to-day tasks. Trust is what we see as taking AI deployment to the next level and being something that really helps an insurer run their business. An underwriter who doesn’t trust the output or the recommendation will route around it, will override it. A carrier that can’t explain the AI-assisted decision to a regulator, a reinsurer, or a policyholder really has a liability and not an aid. So as agentic starts to take on more autonomous work, was it fast really doesn’t concern them. Is it, is it something that I can stand behind and explain to the people that I need to explain things to? And that’s how we’ve really deployed AI across the organizations we work with. It’s showing the results to the customer, and some of the results that we’re seeing right away are 98%+ accuracy on submissions. And that’s an accuracy percentage that’s a real time saver. If we look at submission accuracy that is 95% or lower, that’s something that you’re really going to have to go through each submission anyways to validate the results. Our insurers are now actually processing 67% of their submissions through AI, with an average time savings of 15 to 20 minutes per. That’s really beyond validation and theorized savings. It’s where underwriters are actually trusting the output and delivering value to the organization.

[00:02:22] Paul Lucas: Now, of course, we’ve seen industry discussion move as well, haven’t we? It’s gone from generative to agentic capabilities. So what foundations need to be in place for agentic AI to reliably support insurance decision-making?

[00:02:37] Cameron Scott: Agentic AI is a much higher bar, not one that you can clear with a better LLM, a better model alone. The foundations that we value most are data

foundations and insurance domain expertise. So for data foundations, we’re talking about clean, well-structured, well-governed data that you can rely on and trust. Everybody has heard of, you know, garbage in, garbage out. So it’s really making sure that your data is in a position that you can layer AI on top of it. When we look at deep insurance domain expertise being built into the AI, it’s really not a sequence of developing AI and looking at AI and then layering on insurance expertise after. It’s making sure that the insurance expertise is built in throughout the process and throughout the deployment of AI.

[00:03:26] Paul Lucas: You mentioned, of course, that ISI’s AI in production has seen some strong results. So I’m hoping you could tell us a little bit more, though. What have those deployments actually taught you from a change management perspective in terms of making AI effective in daily underwriting work?

[00:03:42] Cameron Scott: There’s still a lot of fear around AI that we hear about every day. Making sure that you get ahead of the change management problem has been critical. Our deployments that have gone well are the ones where AI was introduced as a digital companion to the underwriter, to the user, as opposed to something that replaces their job or their responsibilities entirely. It’s something that can handle repetitive, lower judgment tasks, service information faster, and really act as a support to the underwriter. That is where we see increases of adoption, increases of trust occurring. Practically, that meant rolling out AI in a way that we have high amounts of training transparency around the content that is produced and the recommendations that are generated, and giving underwriters an easy way to question it and make sure that the AI is always validated and governed throughout the process.

[00:04:35] Paul Lucas: So building on that experience then, what agentic workflows does ISI see as perhaps being the, the next big thing that the industry should be looking at?

[00:04:44] Cameron Scott: Yeah, we at ISI would actually push back a little bit on the next big thing and rather think about what’s the next big thing for each insurer, each carrier, each reinsurer. Our experience has taught us that there’s not one universal agentic workflow that transforms every insurer in the same way. An auto, a personal autos carrier book of business may have a different operational bottleneck than a specialty carrier or than a large commercial carrier. We really value in determining what is the highest place that AI can be deployed and allow insurers to deploy their own agents in those places or work with ISI to deploy the agent to tackle those high-value business cases. So what we really aim for is configurability of agents, the ability to train agents on specific specific carrier’s datasets to tackle those pain points, whether it be submission triage, renewals, endorsements, first notice of loss, whatever is really causing their business the most time savings, or what could provide the most time savings is going to be that next big thing for the specific insurer. It’s giving them the ability to point Agentic AI to whatever is their biggest operational challenge. And make sure that they have the tools to govern it and test it.

[00:06:02] Paul Lucas: Yeah, absolutely. Sounds like the right approach, without a doubt. Cameron, huge thanks for your time today. And of course, thanks to you for watching. We’ll see you next time right here on Insurance Business TV.