The Emergence of AI and Its Impact on Business Consulting

A senior partner at a large firm was asked, some years ago, what his associates actually did in their first three years. His answer was disarmingly honest: they read documents, built spreadsheets, and made slides look expensive. The judgement, he said, came later, and from fewer people than the org chart suggested.

That answer has aged badly. Not because it was wrong, but because the work he described is precisely the work that machine intelligence now does quickly, cheaply, and without complaint. Reading a two-hundred-page detailed project report and summarising its risks used to be a week of an analyst's time. It is now an afternoon, most of which is spent checking the output.

So what remains of consulting when the analytical middle collapses?

What the machines took, and what they did not

Begin with what has genuinely changed. Document review, first-draft financial models, market scans, comparator studies, the tidy synthesis of publicly available information: all of this has become fast. A client in Srinagar or Gurugram can produce a passable industry overview before their consultant's flight lands. Firms that priced this work by the hour are discovering that the hour is no longer scarce.

What has not changed is harder to see because it was always less visible. Nobody ever hired a consultant purely for information. They hired for three quieter things: someone to structure a decision the organisation was avoiding, someone to say an unwelcome thing to a board that would not hear it from an insider, and someone to absorb part of the risk of being wrong. A language model does none of these. It cannot read the silence in a room when a promoter's son is presented with succession data. It cannot decide which of a district collector's five priorities is real.

The Indian context sharpens this. A great deal of consulting demand here comes from institutions where the constraint is not analysis but authorising environment. A public sector undertaking often knows its plant is inefficient; the question is how to sequence change past a board, a ministry, and a union. AI shortens the diagnostic. It does almost nothing to the negotiation.

The uncomfortable part for consultants

Here is the argument few firms want printed. If a third of your fee was quietly funded by junior analytical labour, your margin is now subsidised by a capability your client also owns. Clients have noticed. Procurement teams in PSUs and larger NGOs are beginning to ask why a diagnostic phase costs what it did in 2022. They are right to ask.

The honest responses are limited. Either the diagnostic gets cheaper and faster, and the fee shifts towards design and implementation, or the firm keeps its old pricing and gradually loses the work to smaller outfits that never carried that overhead. There is a genuine opening here for regional consultancies. A six-person practice with good tooling can now match the analytical throughput that once required sixty. What it must supply itself is judgement, sector memory, and the willingness to stand behind a recommendation after the invoice clears.

There is also a new category of work: cleaning up after unsupervised machine output. We have already seen strategy notes with confident, invented citations, and financial projections resting on a hallucinated policy provision. Verification is becoming billable. That is an odd outcome, but a real one.

What clients should now expect

For an organisation buying consulting in this environment, a few expectations are reasonable.

• Diagnostics should be faster and cheaper than they were three years ago. If they are not, ask why.
• The consultant's time should concentrate on interpretation, sequencing, and stakeholder work, not on producing summaries you could generate internally.
• Any AI-assisted analysis should be declared, and someone with a name should have verified it.
• The fee structure should reward outcomes and implementation support, not document weight.

None of this diminishes the profession. If anything, it restores an older idea of it. The consultant as counsellor predates the consultant as slide factory, and the technology, by stripping away the factory, has made the counsel visible again. Firms that were always selling judgement will find this decade clarifying. Firms that were selling formatted information will find it brief.

The emergence of AI, then, is less an invasion than an audit. It has gone through the profession's books and marked which lines were value and which were ritual. The ritual is ending. The value was always rarer than the industry admitted, and it is now, finally, the only thing on sale.