The risk shift in modernisation: how AI is changing the ceiling for insurers and brokers

For much of the last decade, modernisation across the London Market has followed a pragmatic path. Insurers and brokers have invested in improving user journeys, adding workflow layers, and integrating systems where friction was most visible. These changes have helped teams work faster and more comfortably, without disrupting established underwriting models or market practices.

Nevertheless, the risk landscape is shifting, as are the expectations placed on technology.

Artificial intelligence is no longer a future capability. It is increasingly being introduced into underwriting and broking workflows to support document handling, triage, analysis, and decision support. As that happens, many organisations are reaching a point where surface upgrades stop paying back.

This is the modernisation ceiling, and AI is bringing it into focus faster than expected.

AI raises the bar on risk, not just efficiency 

AI is often discussed in terms of productivity: faster turnaround, reduced manual effort, and improved throughput. However, in a regulated, risk centric market like London, AI changes something more fundamental: how risk is created, assessed, controlled, and evidenced within workflows. 

Once AI is used to influence underwriting or placement, organisations must be able to answer harder questions: 

1. What data informed this outcome? 
2. How consistent is the decision across similar risks? 
3. Can the rationale be explained and evidenced? 
4. Who is accountable when AI supported decisions go wrong? 

These questions expose whether modernisation has strengthened the foundations of the workflow or simply improved what sits on top. 

Incremental improvements hold only until AI introduces greater scale, scrutiny, and change by demanding:  

• data consistency at source  

• decision logic that can be tested and explained  

• controls that scale as automation increases  

• change processes that protect stability while enabling innovation 

When those foundations aren’t in place, risk doesn’t disappear it simply moves.  

Underwriters and brokers spend time validating AI outputs, managing exceptions, reconciling discrepancies, and compensating for gaps in data or logic. Governance overhead grows, confidence drops and innovation slows down. At that point, adding more tools no longer improves outcomes, it increases operational risk. 

For underwriters, the promise of AI is compelling: less time extracting information and more time applying judgement. However, that balance only holds if workflows are designed to support judgement rather than obscure it. 

Where foundations are weak: 

• AI recommendations require manual cross-checking 

• decision rationale lives outside the system  

• responsibility for outcomes becomes blurred 

In a market where portfolio risk, regulatory scrutiny, and audit expectations continue to rise, that ambiguity is itself a risk. AI cannot compensate for workflows where risk decisions are hidden, undocumented, or dependent on individual memory. 

What brokers experience and why it matters to carriers 

Brokers face the same tension from a different angle. AI can help reduce administrative burden and accelerate client service, but only if submissions, endorsements, and renewals flow cleanly into carrier systems. 

The result from poorly aligned workflows equals more back and forth, repeated clarification, duplicated effort and placement inefficiency. 

In the London Market, broker and insurer workflows are tightly coupled. Improving one side without addressing the other simply displaces risk across the chain. 

The original promise of modernisation was efficiency, but the emerging requirement is the ability of underwriting and broking workflows to absorb change, scrutiny, and automation without creating new risk or losing control. 

The introduction of AI is reshaping technology’s role in the market, moving it beyond systems of record towards reasoning, judgement support, and continuous change. This requires solid foundations, without this, AI adoption hits a ceiling quickly and increases exposure rather than reducing it.

Five questions that reveal pressure points in underwriting workflows

1.  Can we explain why a decision was reached, not just what the outcome was? 
2. Is our data structured and governed enough to support automation at scale? 
3. Are underwriting and broking rules explicitly expressed or embedded in habit? 
4. Can we change workflows frequently without increasing operational risk? 
5. Do we know who is accountable when AI supports a risk decision? 

Having trouble answering any of the above? See how Verisk is helping the market scale without adding more risk.