AI won’t kill the broker, it’ll kill the boring one(s)..
Broker Expo 2026 ran this year under a headline built to provoke: AI will kill off lazy brokers. It's a good line. It is also, on the evidence from the floor, not really about AI at all.
Charlie Hicks, managing director of EIC Insurance Services, told the show that old-fashioned broking still has “a massive place” in today's market, and that doing it well remains one of the clearest ways a firm stands out from its competitors. His advice wasn't to resist the tools. It was the opposite: “Absolutely use technology to help efficiencies, but don't be afraid of getting out there and looking someone in the eyes.”
That sentence is worth sitting with, because most principals reading the “lazy brokers” headline will assume it means something about adoption speed: who's using Copilot, who's still pasting data between systems by hand. Hicks wasn't talking about speed. He was talking about where the time goes once you've freed it up.
On the same stage, in the session on emerging risks, InsureSafeLine's Ayush Kanungo made a related point from a different angle: brokers should be reaching out to clients proactively, not waiting for a renewal date or a claim to force the conversation. That's not a technology recommendation either. It's a description of what a broker with spare capacity actually does with it, if they're any good at the job.
Put the two together and the real argument behind the Broker Expo headline comes into focus. “Lazy” was never going to mean slow to adopt AI. Adoption is already close to universal across the market; nobody at a 2026 broking conference is arguing against it. “Lazy” means automating the admin and then doing nothing different with the hours it bought you. Same call volume, same reactive posture, same renewal pack finished three weeks before expiry out of habit rather than need, just produced slightly faster by a tool that cost money to install.
This is precisely the trap the operating model this publication argues for exists to avoid. Automate what doesn't require judgement: re-keying, chasing documents, drafting standard correspondence, processing bordereaux, the renewal pack itself. That part isn't controversial anymore; most of the market agrees with it in principle, even where execution still varies wildly. The part that keeps getting skipped is the second half of the instruction. The hours automation frees up are supposed to go to trust, complex advice, negotiation and relationship-building, not back into the same throughput machine with a faster engine fitted.
The reason that second half keeps being skipped isn't really laziness. It's that most firms never did the harder work first: deciding, as a strategic matter, where AI's value actually sits in their own business. Adoption across the market has been operational almost by default, driven from the bottom rather than set from the top. A tool gets trialled on drafting, another on triage, a third on bordereaux, each decision made at the level of “can this save someone an hour,” with nobody in the room asking the prior question: which parts of this firm's advice, relationships and judgement are the ones actually worth protecting and building the rest of the business around. Experimentation is cheap, visible, and easy to report on at the next board meeting. Deciding where the real value sits, and designing the automation programme to protect and compound it, is slower and much easier to skip.
Hicks's “look someone in the eyes” line and Kanungo's proactive-contact point are both describing what that second half looks like in practice, from people who actually run books, not from a slide. Neither framed it as a defence of the old way against the new way. Both framed AI as the thing that makes the old way affordable at scale again, for firms disciplined enough to spend the saved time on it rather than quietly pocket it as extra capacity.
That's a useful, uncomfortable test for any principal who has already rolled out AI tools this year. Pull the calendar data for the top three client-facing people in the business, going back one quarter. Has AI actually bought them more hours in front of clients, used proactively, the way Kanungo describes? Or has the same number of client hours simply been joined by more files processed in the gaps, with the saving banked as extra volume instead of spent on better advice? Most principals won't have a ready answer, and that's really the point. The test only works as a check on execution if someone had already decided, before the tools went in, what the freed-up hours were for. Absent that decision, this isn't a laziness problem. It's a sequencing problem: strategy was supposed to come before the automation, not get discovered afterward by totting up a quarter's diary entries.
If the calendar data comes back as more files processed with the same client hours, the firm hasn't automated the wrong things. Bordereaux, re-keying and renewal drafting are still exactly the right places to start. What's missing is the discipline to protect what the automation was meant to buy back, and the strategic clarity to have named that purpose before the first tool was switched on. A broker who uses AI to process more renewals at the same shallow depth of advice has built a faster version of a business that was already losing ground. A broker who uses it to have more of the conversations Hicks and Kanungo are describing has actually changed what the firm is, because somebody decided that was the point.
Broker Expo put real names and a live stage behind a distinction this publication has been making in the abstract for months. It's worth taking seriously, because the version of “AI will kill off lazy brokers” that's actually true was never about who adopts fastest. It's about who worked out, ahead of the tools arriving, what the time was actually for, and who's still finding that out by trial and error, one experiment at a time.
Source: Insurance Age — “Broker Expo 2026: AI will kill off lazy brokers” (Rosie Simms) — published 9 October 2026; source for the Hicks and Kanungo quotes used above.