By Ravindra Patki
For thirty-odd years, the dominant product an Indian business bought when it wanted external strategic help was the same, regardless of who it hired: a team that would study the problem, benchmark it against other companies, identify best practices, and return with a recommendation. This required a two-tiered effort: junior analysts who did the study and data crunching, working under the direction of an experienced partner. The customer paid for the total effort, with a premium tied to the brand of the firm. On paper, artificial intelligence (AI) should be able to significantly reduce the back-end effort of analysts on a project and invert this pyramid.
The early evidence tells a more complicated story. Several firms have cut analyst headcount after investing heavily in AI tools, but the same period has also produced a string of embarrassing incidents. Recently, a well-known firm withdrew an AI-authored report on agentic AI after an investigation found that most of its citations were fabricated. Another major firm had to refund the Australian government after AI-generated content, complete with invented citations, made it into a taxpayer-funded report. Such incidents have become more frequent now.
The pattern points to the real economics of the shift. Generating a study or a report has become fast and cheap. Verifying that its claims, data, and citations are actually true has not. Fact-checking AI output at the criteria a paying client expects is slow, skilled work, and it does not disappear just because the first draft appeared in minutes.
One possible explanation is that analyst workloads have definitely gone up following recruitment freezes in all these firms. AI has shifted their role from producing content to auditing it, which is a harder and less teachable skill. On top of that, before the adoption of AI, there was no need for a dedicated fact-checking process because the sources were considered authentic to begin with. So, this new working environment of tight deadlines and excessive workloads could have led to these oversights.
There is another cost that receives even less attention. Token usage and associated cost appear to be ballooning in such firms due to the lack of guardrails governing AI usage. People can use AI in any way they choose, generating non-usable outputs while still consuming tokens. As a result, token budgets seem to be increasing in an uncontrolled manner in many firms, which could potentially nullify the expected manpower savings.
Put together, token economics and the cost of verification may not produce any net reduction in what it costs to deliver a trustworthy study. That, more than the headcount story, is the part of this shift getting the least attention.
None of this removes the need for the partner. The ability to contextualise a problem and ask the right question, drawing on cross-industry judgement built over years, is not something AI replicates.
What is changing is who can afford a viewpoint at all. Cheap AI-assisted advice, a first-pass marketing plan, a market-entry framework, and access to best-practice thinking are now within reach of small firms that could never have paid for a top-tier engagement. It will lack the tacit depth an experienced partner brings, but it is a real improvement over the alternative, which for most of these small firms was nothing. Larger organisations are a different case. They will keep hiring consulting firms because they need accountability for the recommendation and buy-in across stakeholders who each have their own incentives, and that is a human negotiation, not an information problem AI can solve. So, the total demand for consulting judgement is unlikely to shrink; what shifts is which firms capture it, and at what price.
That split falls roughly along a line that already existed before AI: firms selling the recommendation itself are exposed, more so because of the changes they are making in anticipation of the disruption, while firms selling implementation are largely not. My own firm sits in the second category; our fee is tied to implementation and outcomes rather than the recommendation, so the direct commercial threat is limited. There is real upside too: AI is useful for building rapid prototypes that make a recommendation concrete for a client, which shortens the gap between advice and action. This is about using AI to enhance our offerings rather than to re-engineer the current method of delivering them.
As for AI use cases being sold as standalone offerings, these have not gained much traction in the corporate world so far. This is a market still finding its shape. Despite the scale of investment going into it, there are not yet many case studies showing measurable improvement in outcomes, and that gap between spend and demonstrated value is worth watching. The basic issue with AI adoption for businesses is all about the accuracy and adequacy of data for training models with company-specific use cases, data security, and hallucination. My expectation is that this market will evolve into some kind of outcome-based pricing rather than token-based fees, once these problems are addressed.
Underneath all of this, the hardest part of consulting has not changed. Deep problem-solving requires understanding context, telling apart core problems from visible symptoms, drawing on cross-domain knowledge that is dependent on observations, and deciphering insights despite bad data, and reasoning from first principles to a causal explanation. Innovative problem solving is not a prompt away from the solution.
Implementation of solutions also requires navigating the emotions of the people who have to live with the solution and getting them to adopt it. None of that is within reach of current AI, and firms built around that kind of work, the implementation-focused firms, will stay largely insulated from this disruption.
Firms selling recommendations as their main revenue product face a transition issue, not because the need for their judgement is going away, but because the economics of producing and defending that judgement have shifted under them faster due to what I believe is a self-imposed crisis. In this phase, the ones that keep report quality high through this transition should hold their position. The ones where clients start noticing the drop in rigour are the ones that will lose market share, and probably deserve to.
The author is Managing Partner, Vector Consulting Group
Disclaimer: The views expressed are the author’s own and do not reflect the official policy or position of Financial Express.
