AI in insurance is no longer hypothetical. When people say “AI is going to change insurance,” it can sound like the “blockchain will change everything” talk from a few years back, but this time the shift is already showing up in underwriting, claims and live books of business.
Except this time it’s not hypothetical. Underwriting cycles that used to take three days are now taking three minutes at some carriers. Straight-through processing rates have jumped from the 10-15% range to 70-90% in some lines. Claims that used to sit in a queue for weeks are resolving 75% faster with meaningfully lower costs. This isn’t a pilot project buried in someone’s innovation lab anymore. It’s showing up in live books of business, right now, in 2026.
Insurance is one of the most interesting industries to watch at the moment, because the pressure is coming from every direction at once: carriers moving faster, clients expecting more, regulators watching closely, and a talent pool that’s stretched thin. Here’s what I think brokers and insurance companies should be paying attention to before the market sorts itself into winners and everyone else.
A quick read on where things stand
The role of the broker is shifting from transactional to advisory. It used to be: gather information, go to market, get quotes, present options, bind coverage, repeat next year. AI is compressing the transactional parts of that cycle and pulling the value of the broker toward the parts that are harder to automate: reading a client’s risk profile, spotting the coverage gap nobody flagged, making the judgment call in a complex placement. The firms already using AI for benchmarking, coverage design and portfolio analysis aren’t cutting headcount so much as freeing their best people to actually advise, more often, with better information.
On the carrier side, underwriting is moving from a once-a-year snapshot to something closer to continuous risk assessment, fed by real-time data. Fraud detection is sharper. Quote turnaround is becoming a genuine competitive factor. In a soft market, the insurer who comes back first with an accurate number wins a disproportionate share of the business.
None of this means brokers or underwriters become obsolete. It means the ones who adapt compound their advantage, and the ones who don’t fall further behind every quarter.
Five things I’d be thinking about right now
1. Your data is the actual bottleneck, not the AI. Almost every broking and carrier conversation I have eventually ends up here. The AI tools themselves are increasingly commoditised; what separates the firms pulling ahead is whether their data is even usable. Information scattered across legacy systems, in inconsistent formats, with gaps nobody’s owned up to, is what quietly kills most AI initiatives before they start. Before evaluating another tool, it’s worth getting honest about whether your data can actually support one.
2. Speed is now a client-facing promise and not a just back-office metric. Quote-to-bind time used to be an internal efficiency stat. It’s becoming something clients notice and compare. If a competitor can turn around an accurate quote in minutes and you’re still measuring in days, that gap is visible to the client, and it’s costing renewals and new business, even when the coverage and pricing behind it are just as good.
3. The broker’s value is moving up the chain, so people need to move with it. If AI is handling more of the repetitive placement and benchmarking work, the value-add shifts toward advisory judgment: spotting underinsurance, structuring complex risk, having the conversation a client can’t get from a portal. That’s a different skill set than pushing paperwork through fast, and it’s worth investing in deliberately rather than assuming it develops on its own.
4. Governance and explainability aren’t optional extras. Regulators have already classified underwriting and claims AI as high-risk in some jurisdictions, which means documentation, bias testing, and a human who can explain a decision aren’t nice-to-haves. They’re the cost of entry. Any AI deployed in underwriting or claims decisions needs that oversight built in from day one. Retrofitting governance after a regulator or a client asks hard questions is a much worse position to be in.
5. Adoption is a people problem before it’s a technology problem. This is the one I feel most strongly about, because it’s the failure mode I see most often. Firms roll out AI tools, usage stalls, and everyone quietly blames the software. Almost every time, the real issue is that the people who’d use it day-to-day weren’t trained, weren’t consulted, or don’t trust the output. AI only works when the team using it actually believes in it, and that takes deliberate change management, not just a license and a launch email.
What this means for you.
None of this requires betting the business on AI overnight. It requires an honest audit of where your data, your processes and your people actually stand today, and a clear-eyed view of where the gaps are before a competitor, or a client, finds them for you.
The firms that get ahead over the next few years probably won’t be the ones with the flashiest AI tools. They’ll be the ones that got the fundamentals right: clean data, clear governance, and people who actually trust the systems they’re using.
We hope this was useful. Have a look at the Data Services we offer to help Insurance businesses enable AI.
Author: Colin Telford, COO.
Follow our LinkedIn company page to stay up to date with all our new blogs!
