Practice 03
AI in the Deal
The thesis leans on AI, and nobody in the room can tell you whether the position holds for twelve months.
Judgement, not a checklist
The work
Nobody can confidently price AI in a deal, because it is a judgement rather than a measurement. We give you ours, with a number on it.
AI in a deal behaves like a Rubik's cube. Four faces move at once and turning one turns the others: what is running today, what it could become, what competitors and customers are building instead, and how quickly a defensible position becomes a commodity one. A target that looks strong at signing can lose its place a month after close. This practice exists because that call takes judgement, and someone has to put their name to it.
Four faces, turning at once
AI is a Rubik's cube, not a checklist.
Assessed face by face, AI always looks fine. The risk lives in the turns: capability that is real today and commoditised by the first board meeting, a roadmap that assumes a data foundation nobody has built, a customer quietly building the same thing in-house. We hold the four faces together and give you a view, not a matrix.
- Real or demo: what the plan assumes about AI, against what is actually in production
- Provenance and IP ownership of generated code, data lineage and licence exposure
- Replicability: what it would cost a competitor to rebuild it now the tooling is cheap
- Run cost at three times the volume, and the governance that keeps the estate legible after close
AI across the Code to Cash Method
Where AI shows up in each stage.
AI is not a one-off report and it is not a scope line. It runs through all four stages of the Code to Cash Method, and the same team carries the view from first look to exit.
- Judgement over logic. A matrix cannot tell you whether a position holds.
- Every AI finding carries a cost, a timing and an owner.
- Lens: Merger OS™ carries the evidence, so answers are cited back to the target's own material.
01
Assess
Is the AI real or a demo, and what does the value-creation plan assume about it.
02
Diligence
Provenance, IP ownership, data lineage, licence exposure, model risk, replicability and run cost.
03
Integrate
Guardrails, governance and the 100-day plan that keeps the estate legible after close.
04
Optimise
AI enablement through the hold, and an exit narrative a buyer's diligence can survive.