Consulting enters the economy of proof

Generative AI is changing what clients buy from consultants. Two pieces of research, read together, show the scale of it. The first shows that intellectual production, the heart of the traditional model, is speeding up dramatically. The second shows that this speed almost never turns into results on its own. Between the two, a simple shift emerges: clients no longer buy time; they buy proof.

Production speeds up, value moves

In 2023, researchers from Harvard, Wharton and MIT gave eighteen realistic tasks to 758 Boston Consulting Group consultants, with or without GPT-4. On tasks within AI’s capabilities, those who had it completed 12.2% more tasks, 25.1% faster, at markedly higher quality. Those tasks (research, synthesis, drafting a first version) are precisely the ones performed by the junior teams of the pyramid model, billed by the hour.

The study says something even more important. On a complex task lying outside what AI can do, consultants using it were 19% less likely to reach the right answer. The tool speeds up production; it does not supply judgement, and can even lead it astray. What remains, and what commands a price, is therefore what the machine does not do: framing the real problem, arbitrating under uncertainty, bringing an organisation along.

Speed is not enough

The second piece comes from MIT. Its report The GenAI Divide (Project NANDA, 2025), based on interviews with executives, a survey of employees and an analysis of hundreds of deployments, reaches a stark conclusion: only a small minority of generative AI pilots integrated into business processes, around 5%, deliver a measurable impact on results. The rest improve individual productivity without changing the P&L.

The lesson applies to consulting as much as to its clients. Accelerating is not enough: speed still has to be turned into value. A deliverable’s value is no longer measured by its page count, but by how quickly it lets you decide, and then by what actually changes once the decision is made.

Billable time is no longer the unit of account

The business model follows logically. When a growing share of production can be accelerated, billing days becomes hard to defend. Buyers look elsewhere: what the engagement made it possible to decide, deploy and improve. Hence the appeal of phased engagements, where only the first phase commits the client and the following ones open on observed results.

Independents move from plan B to strategic lever

On the market side, large corporations increasingly rely on experienced independent consultants, for targeted expertise, flexibility and operational pragmatism. But the potential remains largely untapped: most organisations mobilise individual profiles, piecemeal, without steering.

What they have not yet integrated is that independents can operate as a structured team, led by a mission lead who carries responsibility for the result. This is the format of large-scale projects, until now reserved for the big firms. And AI is speeding up the change: the faster production becomes, the more experience and judgement make the difference.

What we conclude

These movements converge on the same model: tight teams of experienced consultants, equipped with AI, committed to observable results rather than billable time, with a single counterpart accountable for the engagement end to end.

That is the model we chose to build. Not because research validates it (it came after our convictions), but it is always good to see it looking in the same direction. In an economy of proof, only one thing counts in the end: what has actually changed in your organisation.

Sources

• Fabrizio Dell’Acqua et al., Navigating the Jagged Technological Frontier, Harvard Business School, 2023: papers.ssrn.com/abstract=4573321
• MIT, Project NANDA, The GenAI Divide: State of AI in Business 2025, July 2025.