CB Insights and ACORD just analysed 125 companies tied to the nine biggest insurance brokerages, spanning acquisitions, investments, partnerships and alumni-founded startups from January 2024 through June 2026. Among all that activity, exactly four acquisitions involved AI companies. Gallagher and Hub International, the most active buyers, are acquiring what brokerages have always acquired: established firms with deep client relationships and specialist books. Talent, not technology.
That is a rational strategy if your advantage is scale. But buried in the same report is a sentence that should make every mid-market broker and MGA principal sit up. The gains from hyper-acquisition, the analysts conclude, will likely diminish over time as AI makes it easier for challengers, including smaller agencies and brokerages, to compete with less scale.
Read that again. The report about the nine giants concedes that the moat the giants are spending billions to widen is the one AI erodes.
Scale in broking has always bought three things: market clout with carriers, the ability to fund specialists, and the back-office machinery to process volume. The first survives AI. The second and third are exactly what is being commoditized. A five-person specialist MGA can now run submission triage, document production, policy administration and renewal workflows that used to need a fifty-person operation. Hub is deploying Claude to more than 20,000 employees and Acrisure's hiring data shows a priority to build agentic AI into the placement engine, not just the back office. The tools they are racing to adopt are the same tools available to a firm one thousandth their size, without forty acquired back offices to reconcile first.
That last part matters. A consolidator that has bought 59 agencies has bought 59 ways of processing a policy. The integration debt is the price of the strategy. A small firm running one system of record has none of it, which is why the report's own logic points at challengers.
The second thread in the report is the one with a clock on it. Earnings call mentions of agentic commerce went from 7 in Q2 2025 to 133 in Q2 2026. Spanish insurer Tuio's home insurance quoting through ChatGPT moved markets, and Plymouth Rock, Liberty Mutual and Insurify have since followed with their own apps. The report puts it bluntly: brokerages without a marketing plan for LLMs risk losing business to those that have one.
Marketing plan undersells it. An AI agent shopping on a customer's behalf cannot buy from a broker whose product exists as a PDF wording, a phone number and a five-day turnaround. It can only transact where there is something to transact with: a product defined in software, rated by an engine, bound through an API, paid and documented without a human in the loop. When the buyer is a machine, being machine-readable is distribution.
We saw the human version of this last month. A customer of one of our broker customers insured his $495k launch online in about three minutes and called it the standard other businesses should learn from. The plumbing that produced that three-minute experience, quote to bind to payment to documents to trust accounting with no re-keying, is the same plumbing an AI-driven purchase will need. Firms building it for today's impatient humans are, whether they intend it or not, building it for tomorrow's automated buyers. Firms that are not will be invisible to both.
The survey component of the report is small, 25 practitioners, so treat it as directional. But the direction is consistent with the deal analysis. Around half of respondents picked the relationship-led advisory firm as the operating model that best describes the future for agents and brokers. Only one in five picked the highly-automated enabler of transactions.
Those two answers are not in tension. The advisory future is only affordable if the transactional layer runs itself. Every hour a specialist broker does not spend chasing documents, re-keying schedules or reconciling premium is an hour that can go into the work clients actually pay a specialist for: structuring cover for complex risks, negotiating with carriers, being the trusted adviser 72 percent of respondents remain confident the market still wants. Automation is not the opposite of the relationship model. It is what funds it.
So the sorting question for the next five years is not "do we have an AI strategy", which usually means a pilot and a policy document. It is narrower and harder: can a machine quote, bind and pay for your product without a human touching it? If yes, you are distributable through every channel that is coming, and your people are free to do advisory work. If no, then every dollar the consolidators spend on talent takes share from you today, and every advance in agentic commerce takes share from you tomorrow.
The giants are betting scale plus talent. The report they commissioned says the scale half of that bet is depreciating. For a specialist firm with a modern core and a machine-readable product, that is not a threat. It is the best structural news in a decade. What would it take for your best-selling product to be bound online, end of this year, with no one in your office touching the file?