Make smarter decisions on the AI in insurance market

AI in insurance is leaving the pilot phase and moving into the daily work of underwriters and claims teams.

We've found a lot of interesting signals. For example, in February 2026, AIG's CEO Peter Zaffino said AI now lets the insurer handle much larger volumes of underwriting submissions without adding staff. And in July, Allianz Partners, the group's travel and assistance arm, said it would cut 1,500 to 1,800 jobs in Europe, mostly in call centers, with its CEO naming AI as the reason. Large insurers are no longer testing AI on the side. They are reorganizing work around it.

In this section, we look at which insurance jobs AI takes on first, why underwriting and claims moved before pricing, and what these changes mean for the vendors selling to insurers.

Underwriters spend much of their day reading documents instead of judging risk, claims teams drown in paperwork after every storm, and fraud quietly adds to every premium.

We compare six kinds of AI insurance vendors, from underwriting and pricing tools to claims, fraud detection, distribution and policy service. Their buyers range from large carriers and MGAs to brokers and claims administrators. For each, we rank the pains that hurt most, like submissions nobody has time to read, slow first contact after a loss and claims paid out too generously.

This shows which insurance tasks a carrier will pay to hand to AI, and which ones it wants a person to keep signing off.

Annual licenses, fees per submission or per claim, shares of the savings found, and revenue splits with an MGA: AI insurance vendors get paid in very different ways.

We line up who pays for each model and where the money goes. A carrier pays for underwriting tools out of its IT budget and buys slowly. A claims tool paid per file grows with volume, but a calm year means fewer claims. Fraud and leakage tools can charge on savings, which is easy to sell and hard to prove. And every vendor pays for integration work with old core systems.

You'll see which models earn more as insurers use them more, and which ones stay stuck in a long list of pilots.

Insurers have plenty of reasons to use more AI, and they don't all carry the same weight. This section sorts the strong ones from the weak.

The obvious drivers are more submissions per underwriter, experienced staff retiring faster than they can be replaced, and bigger weather losses that flood claims teams. The quieter ones reinforce each other. Each claim an AI handles adds data to price the next policy. A carrier that cuts its costs pushes rivals to follow. And once underwriters trust a tool on one line of business, they ask for it on the next.

Before building or investing, you should know what makes an insurer trust AI with a price or a payout, and what keeps regulators comfortable once it does.

Shift Technology and Tractable come up first when insurers talk about AI, but they cover only a slice of the market.

Some of the most interesting vendors are names the public never hears. One kind only reads loss runs and broker emails for commercial underwriters. Others estimate roof damage from aerial images, flag workers' comp claims likely to drag on, or answer policyholders' calls at night for small mutual insurers.

Few people notice them, because most sell to a handful of carriers under strict confidentiality, and many run inside a core system or a claims administrator's platform. A vendor can process thousands of claims a day and never appear in a press release.

Below, you'll find the market grids and ecosystem maps we've built to show which insurance tasks each company automates, and where new players are getting in.

Federato, hyperexponential, Sixfold… a few AI vendors have become part of how insurers price and write risk.

Federato gives underwriters one screen to sort submissions by how well they fit the insurer's risk appetite, and raised $100M in November 2025. Hyperexponential gives pricing actuaries software to build and update their rating models. Sixfold, younger, reads submission files and writes a first risk summary for the underwriter. Each found a clear job inside underwriting, though none has yet proved it can win across every line and every carrier.

We'll look at how these teams picked one underwriting task, proved it on a few carriers' real books, and grew from there without asking insurers to replace their core systems.