Rent
Generic capability you should buy cheaply and never build identity around. Raw intelligence, common workflows, anything the ecosystem already does well. Renting here is the correct answer, not a failure.
AI Strategy · Framework 02
AI market forces
The margin pool is moving from generic intelligence to contexted outcomes. That single sentence explains most of what the large players are doing, and it tells a normal business where its own value is going to sit.
The thesis
The AI market is not settling into neat layers. It is compressing layers wherever the work is generic, and hardening them wherever the context is proprietary, regulated or commercially sensitive.
Where the eight shifts tell you where value is moving, this tells you why it moves, and what your firm should rent, protect or build.
The three moves
Rent, protect, compound
Every firm has three jobs in this market. Clarity comes from naming which one applies where.
Protect
Proprietary context that is exposed: customer trust, pricing logic, operational judgement, regulated decisions, historic outcomes. Protect it by being deliberate about what the model layer gets to see.
Compound
Context you can grow faster than competitors: the data exhaust of your real work, linked to real outcomes, captured on purpose and fed back in. The only move that builds a durable lead.
Where value sits
Context times friction
Hold two dimensions at once: how generic or proprietary the context is, and how much friction protects it. Regulation, integration, switching cost, compliance, trust.
Generic context, low friction
Open and cheap models win. There is nothing to defend.
Generic context, high friction
Platforms defend for a while on workflow and compliance, but the ground is moving under them.
Proprietary context, low friction
Yours, but exposed. The model layer will try to absorb it through everyday usage.
Proprietary context, high friction
Where margin lives. Context only you hold, protected by regulation, trust, integration or switching cost.
The diagnostic
Five questions that locate your firm
Answer them honestly. A made-up moat is worse than no moat.
- Exposure. Which of the things we sell could a capable model already do without our specific context?
- Unique context. What do we know, hold or see that the ecosystem around us cannot easily learn?
- Leakage. Where are we at risk of handing our proprietary context to a model or platform through everyday use?
- Compounding. Which context could we capture and grow faster than our competitors, if we set out to?
- Trust. Where could we make ourselves part of the trusted infrastructure of our market, rather than an optional layer?
Get the full framework
The full write-up on the forces reshaping where firms make money, the context-times-friction map, the flywheel, trust as a moat, and the interview prompt that turns it into a board-ready read on what your firm should defend or build.
- The full framework write-up
- The tailoring interview prompt
Free. Goes straight to David. We only use your email to send the link and reply, no list and no spam.