Anthropic: Insurtechs will see AI recast their vendor role

Anthropic panelists seated at Zywave AI Exchange, 2026
From left: Doug Marquis, chief technology officer, at Zywave; Jake Sloan, insurance GTM at Anthropic; and Eoghan Scully, applied AI architect at Anthropic.
Daniel Wolfe

At a time when many insurance companies are stuck trying to squeeze small efficiencies out of AI chatbots, Anthropic, the maker of Claude, is seeing a far bigger change happening on the insurtech front.

Processing Content

It's now possible for some insurance companies to use AI coding tools to replicate a lot of what they get from vendors, or to supplement those systems beyond what the vendors offer directly, according to Eoghan Scully, applied AI architect at Anthropic.

The build-versus-buy conversation has been a huge topic of conversation among insurance companies for decades, but recently, "the boundary between what makes sense to buy, versus what makes sense to build, has shifted hugely toward the build," Scully said. 

Scully, who spoke on a panel at Zywave's AI Exchange last week in New York, said this disruption would not be identical across the industry, as the shift to building fully in-house would still require a substantial investment from firms that decide to go that route.

"Some of the companies with large engineering teams will say that the economics have moved so far that it makes more sense for us to dedicate a few of our engineers and a lot of work or token spend" to build an equivalent of their favorite Software-as-a-Service tool, he said. 

Others will choose to keep the vendor relationship and augment the software by building agentic intelligence into it, Scully added. But many companies will go a different route entirely by using their preferred agentic tool as an interface or "window" into applications provided by a vendor, he said. 

"These applications will remain as providers of that data, as systems of record … I think that middle ground is where most will congregate," Scully said.

There is still a lot of time for this dynamic to play out, and for SaaS providers and other insurtechs to adapt to this change. According to PwC, 77% of financial services leaders don't see a measurable ROI from their AI investments, and just 48% see any time savings on routine work. 

But as their clients catch up, insurtechs can get ahead of this change by embracing their potential role as a tie-in to the AI interface their clients prefer.

"We'd agree that when you think about the system of record, the interaction with those systems from a user perspective is going to change," said Doug Marquis, chief technology officer for Zywave, the insurtech that hosted the panel. 

The important part is the data, and providing access to that data for AI agents to operate, Marquis said. Zywave facilitates that interaction.

"Am I worried that there's an existential threat to our business? Not really, because we build all the things around it that connect everything together," Marquis said.


For reprint and licensing requests for this article, click here.
Insurtech Artificial Intelligence
MORE FROM DIGITAL INSURANCE
Load More