Consulting: preserving the apprenticeship while AI changes the work

Consulting firms are experimenting with AI across research, analysis, drafting and knowledge retrieval. The opportunity is real, but uneven. A Harvard Business School field experiment with consultants found substantial benefits on tasks within the technology's effective range and introduced the idea of a “jagged” frontier, where performance varies by task. The study is evidence about its tested setting, not a promise that all consulting work will improve by the same amount.

The executive challenge is larger than tool adoption. A consulting leader must decide how expertise is formed, how quality is assured, what clients should pay for and which work the firm should stop doing. This calls for judgement about the business model as well as technology.

Follow the learning path

Junior consultants often develop by doing structured but imperfect work: forming hypotheses, checking sources, building analyses, receiving challenge and revising. If a tool produces a plausible first answer, the learner may move faster but see fewer of the steps that make an answer defensible. The firm must redesign supervision so that speed does not become shallow understanding.

A strong leader can specify where AI may assist and where direct reasoning must remain visible. They might require a consultant to explain source quality, alternatives considered and limitations before a recommendation reaches a client. They should also create new learning tasks, not merely remove old ones. If preparation time falls, what work will develop commercial judgement, client listening and the ability to challenge a flawed problem definition?

Reconsider the economics honestly

The traditional relationship between staffing, time and fees may change when some tasks become faster. A leader should be able to discuss pricing, margins, investment in proprietary knowledge and the client's perception of value without pretending that every hour saved becomes profit. Quality review, secure technology, training and rework all have costs. The market may reward a different service model, but the firm needs evidence.

Candidates for a managing-partner or practice-leader role should be asked about a service they redesigned. What changed for the client? Which parts became more repeatable? Which required more senior involvement? How were younger professionals developed? What was measured, and what turned out differently from the initial plan? The questions separate real operating leadership from a presentation about the future of consulting.

The search should also test how the candidate changes incentives. If partners are rewarded only for individual sales, they may be reluctant to contribute methods, data or training that make the whole firm stronger. If productivity gains accrue only to the current engagement, investment in shared capability can be neglected. An incoming leader needs to know which elements of the partnership model they can influence and how they would bring peers into a decision about shared assets.

Clients may also ask what they are buying when parts of the work become faster. A strong leader can explain the value of diagnosis, challenge, implementation and judgement without defending hours for their own sake. The candidate should be able to discuss how transparent use of AI fits the firm's confidentiality commitments and how senior professionals verify work before it becomes client advice. That conversation tests commercial candour as well as technological fluency.

The search must test institutional stewardship

A rainmaker can create immediate revenue and still leave a practice weaker if knowledge, talent and relationships remain concentrated around them. AI increases the importance of deciding which intellectual assets should become institutional and which expertise depends on an individual professional's judgement. The appointment brief should therefore include the health of the talent pipeline and knowledge system alongside commercial growth.

Four questions sharpen the mandate:

• Which parts of the work create distinctive client value, and which are becoming easier to reproduce?
• How will the firm develop judgement if routine analytical steps become faster?
• What quality checks must remain visible to clients and leaders?
• What business-model choices can the incoming leader actually make?

The strongest consulting leader uses AI to improve the work while protecting the process that creates excellent professionals. A search that measures only current revenue or AI enthusiasm will miss that dual responsibility.