"The AI capability is working and the numbers look good. What I cannot tell the board is whether any of it is still an advantage in three years."
The common finding is that the advantage is real and shorter-lived than assumed, and that what protects it is the data and the workflow rather than the model. It is a less exciting answer than the market rewards, and worth having before an investor asks it for you. This is the analysis behind a no-go I recommended on a pre-IPO round, where the price assumed a lead unlikely to last the hold.
Most AI advantage decomposes into three things: a model, the data it runs on, and the workflow it sits inside. Only two of those are usually yours. The first job is establishing which part of the advantage would survive a competitor buying the same model licence tomorrow, because that is the part with a defensible claim attached.
The economics get run forward on the assumption the capability becomes ordinary. Inference and data preparation costs behave differently as volume grows, and margin built on a capability gap behaves differently from margin built on switching costs or proprietary data. The question is which one you have.
A stated view on how long the advantage holds and what erodes it, with the assumptions written down so they can be argued with. Then what would extend it. That is normally a data or workflow position, not a model one, and it comes with a cost.
A view on whether the advantage is structural or temporary, with the economics behind it, in a form that holds up in front of someone who would prefer a different answer.
The model stays with you and can be re-run as costs move. Inference pricing and capability both move quickly, so last year's answer will not hold.
Where this goes to investors, the horizon and the assumptions behind it are written to be challenged, because that is the first thing a sceptical reader will attack.
In scope: separating the advantage into model, data and workflow, running the economics forward as capability commoditises, a stated horizon, and what would extend it.
Out of scope: technical audit of the models themselves, and deciding which AI use cases to fund in the first place. That is Growth Option Prioritisation.
Fixed price, not a day rate. The scope is agreed up front and the price does not move with it. If the work takes longer than expected, that is my problem rather than yours, which is the right way round.
Quoted after a short call, once I understand what you are actually dealing with. Invoiced half on start and half on delivery.
Most engagements end at delivery. Some clients keep me on a light retainer afterwards to keep the model current and to be available when the board asks something new. That is agreed at the end, not the start.
Where AI does part of the work, I say which part. Some of the analysis and model building uses AI tooling. The judgement, the method and the conclusions are mine, and I will tell you which is which if you ask.