As the insurance industry grapples with the rapid integration of artificial intelligence, Simon Glairy stands at the forefront of the conversation regarding risk management and regulatory compliance. With extensive experience in the Insurtech space, he has witnessed firsthand how the allure of efficiency often outpaces the necessity for oversight. In this discussion, we explore the widening governance gap within the Managing General Agent market, examining the friction between high-speed AI adoption and the rigorous standards required by capacity partners and regulators alike. From the stark differences in readiness between MGAs and Lloyd’s managing agents to the looming 2026 regulatory focus, Glairy provides a deep dive into the operational and commercial realities of a sector undergoing a fundamental digital transformation.
While over 80% of MGAs have already integrated AI into their workflows, barely half have established a formal governance framework. What do you believe is driving this dangerous mismatch between technology adoption and organizational control?
The speed of this shift is truly breathtaking and, frankly, a bit unsettling for those of us focused on risk. We saw generative AI tools achieve mainstream adoption in just a few weeks—a pace that makes the multi-year growth trajectories of platforms like Facebook or Netflix look like they were moving in slow motion. This “frenetic rush” has left many of the 300-plus MGAs in the UK scrambling to keep up, often prioritizing the immediate efficiency gains of AI over the boring but essential work of building guardrails. When more than 80% of these firms are using the tech, but only 52% have a framework to manage it, you can feel the tension in the room; it’s like watching a high-performance engine being installed in a car with no brakes. MGAs are essentially operating in a gold-rush mentality, but without governance, they are exposing themselves to “data poisoning” and hallucinations that could quietly rot their underwriting decisions from the inside out.
Lloyd’s managing agents show a 93% governance rate, creating a massive 41-percentage-point gap compared to the MGA market. How does this discrepancy impact the commercial relationship between coverholders and their capacity providers?
This gap isn’t just a statistical curiosity; it is a major commercial roadblock that is starting to chill relationships across the delegated authority market. When 93% of Lloyd’s managing agents have their AI house in order, they naturally expect the same level of rigor from the partners writing on their paper. If you’re a coverholder without a governance framework, you’re now a “third-party risk” in the eyes of your capacity provider, which can lead to grueling delegated authority reviews or even the loss of capacity altogether. I’ve spoken with executives who feel the weight of this discrepancy deeply, realizing that their lack of internal controls is no longer just an operational oversight but a threat to their very survival. With delegated authority’s market share forecast to exceed 45% by 2027, the pressure to close that 41-point gap is immense because no managing agent wants to be held liable for an MGA’s “black box” decision-making.
Beyond the boardroom, what are the specific operational and liability risks that keep MGA leaders awake at night when they realize their AI tools are running without sufficient oversight?
The nightmares usually revolve around the “ghost in the machine”—those AI hallucinations and inaccurate outputs that can lead to disastrous underwriting or claims decisions. Because MGAs handle a massive volume of sensitive personal data, the fear of a data leakage or an AI-enabled cyber attack is very real and carries a visceral sense of dread. There is also the looming shadow of intellectual property loss and copyright exposure, which can lead to lawsuits that are as expensive as they are reputation-damaging. When you’re managing a portion of the £47 billion in general insurance premiums, the stakes aren’t just theoretical; they are calculated in millions of pounds and the trust of your policyholders. Senior managers are also starting to feel the heat of “accountability risk,” knowing that under current regulations, they are the ones who will have to answer when an unmonitored algorithm goes rogue.
The FCA has indicated it will use existing frameworks like the Consumer Duty to oversee AI, rather than writing new rules. How should MGAs prepare for this “indirect” regulatory scrutiny, especially with 2026 identified as a key focus year?
The regulator is playing a very sophisticated game by relying on the Senior Managers and Certification Regime and the Consumer Duty to enforce AI standards. By naming AI adoption as a formal focus for 2026, the FCA has essentially fired a warning shot across the bow of every firm underwriting a share of that £47 billion premium pool. MGAs need to realize that “no AI-specific rules” does not mean “no consequences”; in fact, it means the regulator will use every existing tool to hold leadership personally accountable for transparency and claims handling. I advise firms to treat 2026 as a hard deadline, because by then, the “honeymoon phase” of AI experimentation will be over, and the expectation of practical guidance on accountability will be the new baseline. You can almost feel the regulatory net tightening, and for the 300+ MGAs in the UK, the time to move from “using AI” to “governing AI” is rapidly running out.
What is your forecast for the MGA market as we approach the 2027 milestone of delegated authority dominance?
I predict a period of “forced maturity” where we will see a significant consolidation of MGAs, driven largely by who can and cannot meet the 93% governance benchmark set by Lloyd’s. As delegated authority’s market share climbs toward that 45% forecast, the “governance-ready” MGAs will attract the lion’s share of capacity, while those stuck in the 52% minority will find themselves increasingly isolated and unable to secure backing. We will also see AI governance move from a checkbox exercise to a competitive advantage, where the ability to prove “clean data” and “hallucination-free” underwriting becomes as valuable as the capacity itself. Ultimately, the market will split into two tiers: those who treated AI as a toy and those who respected it as a powerful, regulated tool that requires a steady, governed hand to steer.
