The canon of existential threats to insurance brokers is well-worn: direct distribution, price-comparison engines, and the occasional insurtech that promises to make middlemen obsolete. Add artificial intelligence to the list—and then take it off. For the global giants that dominate broking and reinsurance intermediation—Marsh McLennan, Aon, WTW and Gallagher—the early evidence suggests AI is less a guillotine than a lever. It automates drudgery, widens advisory scope and, crucially, creates new insurable risks that brokers are uniquely positioned to quantify and place.

Start with the plumbing of the business. Broking produces oceans of unstructured data: emails, PDFs, loss runs, policy wordings and claims notes. Generative AI and newer OCR engines are finally making a dent in this mess. Reinsurance intermediary Gallagher Re points to a basic but costly choke point—submissions that arrive in the “wrong” format and have to be re-keyed. Gallagher Re’s InsurTech report underscores how AI-centric tools are drawing a growing share of scarce venture funding as capital returns to pragmatism.

WTW shows how the line between broker and software house is blurring. Its long-running Radar pricing suite—sold to carriers and used by consulting teams—now bakes in machine-learning techniques and, in its latest update, generative-AI features aimed at accelerating model building, testing and deployment. The business case isn’t theoretical: users report rate-change lead times shrinking from weeks or months to days or even hours, the kind of cycle-time reduction that compounds across a portfolio. Speed to market is not just a carrier advantage; it strengthens a broker’s pitch that it can marshal analytics to secure better terms, faster.

Aon, meanwhile, has pursued digital distribution at the small-commercial end, where AI-driven onboarding, eligibility checks and appetite matching can shave minutes off tasks that used to take days. Its CoverWallet platform—acquired before the current AI boom—has become a laboratory for data-driven quoting and service for micro- and small-business clients, the segment most exposed to disintermediation risk. If AI commoditizes simple covers, incumbents with modern digital storefronts are likely to keep more of that traffic.

The most durable opportunity sits on the advisory side. AI isn’t merely a tool for brokers; it is a new class of client risk to be measured, mitigated and insured. Companies deploying generative models face novel mixes of old perils—IP infringement, data leakage, cyber escalation, discrimination claims and operational outages. Marsh’s analysis sums up an important early finding: generative AI tends to amplify familiar risks rather than invent entirely new ones. That is catnip for an industry that thrives on quantifying known unknowns and negotiating coverage across a fragmented capacity market.

If you want a glimpse of where this goes, watch the market for AI-liability covers. Reporting this autumn indicated that frontier-model developers have been exploring large policies for copyright and other AI-related exposures, with Aon identified by press accounts as a broker involved in arranging capacity. The numbers are fluid and the capacity limited, but the direction is clear: when a new technology creates hard-to-price exposures, the broker that can convene reinsurers, draft workable wordings and monitor loss experience sits at the fulcrum.

None of this immunizes brokers from competitive pressure. In small-ticket commercial and personal lines, AI lowers the friction of direct distribution and embedded insurance. Carriers and MGAs armed with better triage and straight-through processing can meet more demand without a human in the loop. And insurtech investment—though far below 2021 froth—has begun to rebound, with AI-focused firms accounting for a rising share of deals. The technology can cut both ways: the same models that help brokers summarize loss histories can help carriers auto-bind what used to require a phone call.

There are hazards, too, that could boomerang back on intermediaries. Overreliance on “black box” models invites errors-and-omissions claims if a placement recommendation proves ill-founded. Data-handling and privacy commitments in cloud-hosted AI tools must stand up to discovery and regulatory scrutiny; boards will expect their brokers to vet those controls or, at minimum, to explain them plainly. And there is a cultural challenge: AI that speeds drafting and comparison work is useful, but unless firms rewire workflows—deciding what is automated, what is reviewed and what is escalated—the productivity gains will stall, a pattern consultants have seen across financial services.

So is AI a threat or an opportunity?

For the broking oligopoly, mostly the latter. The defensible bits of their franchise—advice on complex risks, deal-making across markets, wordings craftsmanship, and global claims advocacy—are complements to AI, not substitutes. The technology squeezes costs out of back-office grind and opens fee-earning adjacencies in model risk, compliance and cyber. Where AI does threaten margins—in commodity products—leaders are already hedging with direct and embedded platforms of their own.

The bigger strategic risk may be complacency. Scale alone won’t guarantee advantage; it is the quality of a firm’s data, the discipline of its model governance and the speed with which it industrializes use cases that will separate winners from also-rans. On that score, the signs are encouraging. The brokers building real software, wiring AI into frontline work and showing up in the boardroom as risk translators are more likely to see AI enlarge their moat than wash it away.