Insurance
Half your underwriting intelligence retires in fifteen years. None of it is documented. The rate manual stays. The judgment disappears.
This page explains the problem, shows how we solve it, and offers you a way to begin.
The Short Version
Your senior underwriters built their judgment over decades — pattern recognition that lives in intuition, not documentation. Half of them retire in the next fifteen years. Who's capturing what they know?
We use AI-assisted dialogue to surface that implicit knowledge, systematise it, and deliver it as a living knowledge system. Your underwriting expertise becomes infrastructure — trainable, queryable, and independent of any individual.
Why This Matters Now
AI agents are entering insurance. They'll run on whatever you've given them. The question is: will they inherit your underwriters' judgment, or just policies?
Selfhood isn't technological — it's linguistic. We build the identity layer that agents inherit. The humanities entering the AI stack.
The Problem
Why do knowledge initiatives fail? Because the expertise that prices risk and spots fraud isn't data; it's pattern recognition built over decades, and it resists every traditional method of capture.
A senior underwriter looks at a submission and knows something is wrong. They can't explain how. It's pattern recognition from thousands of decisions. Ask them "how do you know?" and you get rationalisation, not the mechanism.
Underwriting intuition here. Claims expertise there. Relationship intelligence somewhere else. No central repository because tacit knowledge can't be stored in databases. It's distributed across ageing human brains.
One quarter of the industry is 55 or older. 400,000+ positions unfilled by the early 2030s. Only 4% of Millennials want insurance careers. The expertise is leaving faster than it can be replaced.
The Solution
We use the spiral: recursive dialogue that moves wide to gather, tight to compress, until tacit knowledge becomes transmissible.
01
We surface what your experts know but can't explain. Recursive conversation, not interviews.
02
Fragments meet. Contradictions surface. A unified intelligence begins to form.
03
Knowledge that doesn't walk out when people do. Searchable, teachable, evolvable.
04
AI agents that think with your institutional intelligence. They speak with your voice.
YOUR EXPERTISE SURVIVES. YOUR INTELLIGENCE COMPOUNDS.
What We Extract
Every insurance business has three layers of tacit knowledge. We surface all of them.
"Is this a good risk?"
The intuition that prices at 2X when the manual says X. The felt sense when something is off. Pattern recognition built from decades of submissions.
"What's actually happening here?"
Which injury patterns suggest exaggeration. Which repair shops pad invoices. Which areas have fraud rings. Detective work disguised as admin.
"Who can we trust, and why?"
The handshake network that determines where business flows. Who trusts whom, built over decades. Worth billions. Documented nowhere.
Entry Points
Different situations need different starting points. Whether you're facing mass retirement, navigating a merger, or building AI capability, we meet you where you are.
Knowledge Transfer
“Extract what cannot be taught”
Your senior underwriters built their judgement over decades: pattern recognition that lives in intuition, not documentation. We extract it, systematise it, and make it transmissible before they retire. The Codex becomes the living repository of everything they know.
View DemoPractice Environments
“Practise against the actual person”
Build forensic profiles of syndicate leads, key brokers, and regulatory interviewers. Turn them into AI agents you can rehearse with: endlessly, privately. Stop practising on live prospects. Know every question before it's asked.
View DemoDocument Architecture
“Rewrite everything from the inside out”
Your underwriting guidelines, broker presentations, and authority matrices accumulated over years: different authors, different moments. They no longer cohere. We rebuild your entire language architecture from a unified centre. Every document becomes a strategic asset.
View DemoAI Capability
“From prompting to partnership”
Your team uses AI like a search engine: “summarise this submission,” “draft a decline letter.” We teach them substrate thinking: building AI containers that hold your underwriting philosophy, speak as your organisation, and surface insights neither human nor AI could reach alone.
View DemoEmergency
“Stop the bleeding, rebuild coherence”
Key underwriter defection. Merger integration chaos. Regulatory investigation. When organisational coherence breaks suddenly, we deploy within 72 hours. Stabilise first, then build the foundation for systematic recovery. Emergency work for emergency situations.
View DemoWhere It's Most Acute
Nowhere is the knowledge crisis more concentrated than in the London market. Knowledge fragments across syndicates, managing agents, brokers, and coverholders. The box culture concentrates tacit expertise in relationships that resist digitisation. There is no central policy or claims database. When leads retire, the knowledge that priced the market disappears.
84
Syndicates
401
Registered brokers
904
Entry-level hires (2023)
400%
Hiring increase needed
What You Build Now Outlives You
The window is 12–24 months. First-movers are already compounding. Thirty years of underwriting judgment can become a knowledge tradition that survives any departure.
operator@deepselves.com