Transforming London Market Underwriting Through Agentic AI

Summary of our recent breakfast briefing on transforming London market underwriting

Context

Ahead of the breakfast briefing, attendees were asked two short questions about their data strategy challenges and current AI deployment. The responses provide a useful snapshot of how London market organisations are currently approaching data, automation and AI.

The feedback suggests the market is actively experimenting with AI tools, but progress is still constrained by data quality, manual processes and legacy operating models. Many organisations are now shifting their focus from experimentation towards building the foundations required for AI to deliver measurable operational impact.

“AI isn’t the hard part. Getting the data, processes and ownership right is.”

---

What the market is experiencing

• Data quality and ownership remain inconsistent across organisations and partners.
• Many underwriting and operational workflows still rely heavily on spreadsheets and manual intervention.
• Important underwriting insight is often trapped in documents rather than structured systems.
• AI tools such as Copilot are being deployed widely, but operational use cases are still early stage.

“Thousands of manual spreadsheet processes still exist across the business.”

Early AI use cases emerging

• Document data extraction from slips and schedules
• Endorsement classification
• New business triage and prioritisation
• Automated decline decisions

Leadership challenge

For many organisations responsibility for enabling these changes is increasingly sitting with COOs and operational leaders. Delivering agentic processes requires changes to data governance, underwriting workflows, technology architecture and organisational accountability.

Key discussion themes

• Moving from risk‑by‑risk decisions towards portfolio-driven underwriting
• Fixing data quality, ownership and standardisation challenges
• Overcoming legacy technology and cultural barriers to automation
• Turning AI experimentation into measurable underwriting and operational outcomes

“AI is just another tool – just a game‑changing one. The real question is how we make it work.”