Every value creation plan now has an AI line

Every PE deck now promises the same things from AI. Operating leverage. Proprietary IP. A moat. Bankable proof in the P&L.

This is what I have learnt doing it from the inside.

Proof in the P&L comes from three disciplines. Pick the right use cases: map the decisions inside your workflows, size each one against an EBITDA driver, and rank them by value and readiness. Execute properly: baseline the decision before you build, then take it into production, not into another pilot. Then run benefit realisation: one business owner per initiative, tracked against the baseline, and reported to the board.

Operating leverage comes from how you automate the workflow. Do not digitise the old process. Map the decisions inside it, codify how your best people make them, then redesign the workflow around them. The system handles the volume, your people handle the exceptions, and autonomy grows as the evidence does. That is when revenue can grow without headcount growing with it.

The IP is the decision logic, not the model. Everyone rents the same models. Your rules, your precedents and your experts judgement, codified and versioned, belong to you. So does the knowledge your people carry in their heads, once it has been captured in a knowledge base the business owns. It stays when they leave, and a buyer can see it.

The moat is compounding. Every decision you build reuses the knowledge and components of the one before. The second is faster than the first. A competitor who starts later starts from zero.

Bankable proof is boring. Adoption, cycle time and exception rates, reported every month before EBITDA moves. Then the line in the accounts. The year-one work nobody celebrates is what produces the year-two returns.

If you cannot trace a saving to a line in the accounts, assume it is not there.

That takes transformation discipline, and experience. That takes the right approach and asking the right questions.