The line between AI consulting and AI implementation, and why it matters for pricing

AI consulting, the strategic work of deciding what to build and why, and AI implementation, the technical work of actually building it, are different skills with different value, and they routinely get sold and priced as one undifferentiated engagement, which creates a quiet incentive problem worth naming explicitly.

When strategy and implementation are priced together as a single time-based engagement, there's a structural incentive to recommend more implementation work than the strategy actually justifies, because the same party is both diagnosing the problem and billing for the cure, a conflict that doesn't require any bad faith to produce a bias.

Separating the two, even informally, produces better outcomes: a strategy phase priced and delivered as its own thing, with a clear recommendation the client could theoretically take to a different implementer, followed by an implementation phase that the client chooses to award, possibly to the same consultant, based on the strategy's merits rather than as an automatic continuation.

This costs some consultants revenue in the short term, since a client is free to take the strategy elsewhere for implementation. It's a more honest structure, and honest structures are, with real consistency, the ones that produce clients who trust the next recommendation enough to come back for it.