The Biggest AI Opportunity for Insurance Brokers Is Not Using AI
Much of the insurance broking market is focused on how AI might improve internal productivity: drafting emails, summarising documents and reducing renewal administration.
Those uses matter. But they may not represent the greatest opportunity.
The more valuable role for brokers could be helping clients understand, manage and transfer the new risks created by AI.
AI is creating interconnected business risks
Recent reports that Nvidia is exploring insurance structures to help support investment in AI infrastructure illustrate the scale of the opportunity. Insurance is potentially becoming part of the financial machinery enabling AI growth—not simply a means of responding to cyber incidents.
The challenge is that “AI risk” does not fit neatly within one insurance product.
A technology-dependent business may face:
- Physical damage to data centres, chips and cooling systems.
- Business interruption following power, network or cloud failure.
- Cyberattack, data loss or misuse of computing capacity.
- Professional or product liability arising from AI-generated outputs.
- Directors’ and officers’ exposures arising from poor governance.
- Dependence on a small number of cloud, model and technology providers.
- Contractual liabilities when an automated service fails.
Clients experience these as connected risks. Brokers, however, often address them through separate policy classes and specialist teams.
The broker organised around insurance products sees only fragments. The broker organised around the client’s business model can see the whole exposure.
From insurance placement to risk architecture
The future broker will need to do more than identify another policy.
Its role will be to help clients distinguish between:
- Risks that can be reduced through better controls.
- Risks that can be transferred contractually.
- Risks that can be insured.
- Risks that must be financed or retained.
- Uncertainties that remain fundamentally speculative.
This is risk architecture rather than product distribution.
It requires brokers to connect conversations involving the client’s finance, technology, risk and operational leaders with the capabilities of insurers and other providers of risk capital.
The practical output might include:
- An AI and technology dependency map.
- Cross-policy coverage and exclusion analysis.
- Plausible loss scenarios.
- A schedule of controls and evidence for underwriters.
- Identification of uninsured or uninsurable exposures.
- A plan covering mitigation, retention and risk transfer.
This is visible, valuable advisory work for which brokers may be able to charge transparently.
The annual renewal is no longer enough
AI systems, permissions, software integrations and technology providers can change throughout the year. A client might move from using AI to assist an employee to allowing an autonomous agent to make decisions or initiate transactions.
That could materially alter its exposure without triggering an insurance review.
Brokers therefore need proportionate change triggers between renewals. These might include a client:
- Giving an AI agent authority to make payments or alter records.
- Introducing AI into customer-facing or safety-critical decisions.
- Changing a critical cloud or model provider.
- Training a model using sensitive or proprietary information.
- Accepting contractual responsibility for automated outputs.
A once-a-year question asking whether the client “uses AI” will not capture these changes.
Brokers must reorganise around client problems
Creating an isolated “AI insurance” desk is unlikely to be sufficient. AI exposures cross property, cyber, professional indemnity, casualty, D&O, crime and business interruption.
Brokers will need multidisciplinary teams capable of examining the client’s operating model as a connected system. Incentives and revenue attribution will also need to support collaboration across traditional product silos.
The opportunity is not to make AI risk sound more mysterious. It is to make a complex and fast-changing environment intelligible enough for clients to manage and insurers to underwrite.
Using AI internally may help brokers work more efficiently. But efficiency alone is not a strategy.
The greater strategic opportunity is to become indispensable to clients whose technology providers and conventional insurance programmes cannot, independently, explain the risks they face.
Questions for broker leaders
- Are we treating AI only as an internal efficiency initiative?
- Can our teams assess a technology-dependent client across every relevant insurance class?
- What changes should trigger a risk review between annual renewals?
- Do we understand clients’ shared cloud, model and software dependencies?
- Which elements of this work should be provided as separately valued risk advice?