If everyone has AI, where does competitive advantage move?
10 — Return to 2026
The future is not the strategy. Our response is.
We are back on a Tuesday in 2026. Nothing in 2029 was a prediction — but every question it raised is already answerable today.
Executive impact cockpit
Illustrative scenarioIf this future arrives…
AI adoption45% — Medium
LowMediumHigh
AI autonomyAgent
AssistantAgentAutonomous
Human involvementMedium
HighMediumException-based
External AI competition50%
LowMediumHigh
Underwriting capacity131
Indexed submissions per underwriter
Cost per submission£134
Fully loaded, illustrative
Processing time17.7 hrs
Submission to indicative quote
Risk visibility59 idx
Portfolio signals surfaced per week
Human referrals36%
Share of cases touched by a human
AI operating exposure29 idx
Decisions dependent on autonomous systems
Governance complexity46 idx
Controls needed to operate safely
Combined ratio95.6%
Pressured by external AI competition
Note the shape of it: pushing adoption and autonomy right improves capacity and speed and simultaneously raises AI exposure and governance complexity. There is no setting where the second column goes away.
Six decisions
What should we do now?
01BuildWhere should we build proprietary AI capability?Buy the commodity layer. Build only where our data, our risk view or our workflow makes the result impossible to replicate.
02DataWhat data could become a competitive advantage?Survey reports, claims narratives, declined submissions, broker interactions. Most of it is unstructured and currently treated as exhaust.
03WorkWhich workflows should be redesigned rather than simply automated?Automating a 2026 process gives a faster 2026 process. Renewal, referral and claims triage deserve redesign, not acceleration.
04TalentWhat skills will matter when AI does more execution?Challenge, judgement under uncertainty, negotiation, portfolio thinking, and the ability to supervise systems that work faster than we do.
05GovernanceHow do we safely move from AI assistance to AI action?Autonomy earned in bands, with evidence. Governance designed as part of the operating model, not bolted on after the pilot succeeds.
06DifferentiationIf competitors get the same models, what will make us different?The uncomfortable question. If the honest answer is 'nothing yet', that is the strategic agenda.
Final question
If our competitors have access to the same AI…
What should we own?
What should we redesign?
What should we stop doing?
What should we become exceptionally good at?
Strategic canvas
So… what should we start changing now?
Type in the room. Saved in this browser so the notes survive a refresh mid-session.
StartWhat do we begin in the next 90 days?
StopWhat are we doing only because we always have?
ReinventWhat should be redesigned, not automated?