The 2026 AI capability map for financial firms.
Where the firm has to be by end of 2026, and where it is fine to still be pending. The honest capability map.
By end of 2026, three tiers of AI capability will separate serious financial firms from unserious ones. Below is the map we brief boards with — what should already be in place, what should be shipping this year, and what is still frontier.
Table stakes (should already exist)
- Private RAG on internal documents (Firm Memory v1).
- One internal copilot with provenance.
- Model policy — at least one classifier deciding private vs frontier.
- Basic evals on the top two use cases.
Shipping this year (differentiator)
- A live Model Router across at least two providers.
- Two or more agent-run workflows with audit.
- A Skills registry with ten or more versioned skills.
- Human-in-the-loop escalation policies.
Still frontier (fine to be pending)
- Multi-agent orchestration across desks.
- Continuous fine-tuning on firm signal.
- Cross-firm federated evals.
Where to start
Every KJ Capital engagement begins with a two-week £15k AI Diagnostic. We map your existing data estate, model your firm's OS across the six layers above, cost the build, name the first three skills to ship, and hand you a plan your board can act on. It is deliberately priced to be a rounding error against the value it unlocks.
FAQ
How do we benchmark ourselves?
Start with the free 5-minute Readiness Score. It maps directly onto this capability map.
Start with the £15k AI Diagnostic
Two weeks, one costed OS blueprint, three named skills to ship first.