When Analysis Becomes Abundant, What Becomes Scarce in Consulting?

It used to take them four hours; now it takes five minutes.

Xavier Aymonod, Partner at Talan and former Partner at Roland Berger, uses this example to capture how AI is already reshaping consultants’ work.

Yet he immediately adds that part of the time saved should still be spent reviewing, checking and challenging the result. This productivity gain therefore reveals a paradox: AI is increasing the value of human judgment at the very moment it is automating some of the tasks through which that judgment used to develop.

When analysis is no longer scarce
Research, translation, synthesis, structuring and initial benchmarking are all tasks that can now be accelerated by AI. Brynjolfsson, Li and Raymond (2025), studying 5,172 customer-support agents, find an average productivity increase of 15%, with the strongest gains among less experienced workers. AI does not eliminate expertise, but it makes some capabilities that were previously linked to experience much more accessible.
In consulting, this shifts the basis of differentiation. As Aymonod puts it, “everyone will have the same AI.” If access to a first high-quality analysis becomes widespread, saying that “soft skills” will matter more is no longer enough. The real question is not simply who communicates better, but who knows what to do with an answer produced almost instantly by a machine. This is where judgment becomes critical.

Judgment is not a single skill
In consulting, judgment can be understood through three practical questions. First, validity judgment: is this answer correct, complete and well supported? Second, contextual judgment: does it make sense for this client, in this situation, at this particular moment? Third, decision judgment: what should actually be recommended, with which trade-offs and level of risk?
This distinction matters because an AI-generated answer can sound convincing without being reliable. Dell’Acqua et al. (2026), in an experiment involving consulting tasks, show that AI can improve both speed and quality on some tasks, while reducing performance when users operate beyond its “jagged technological frontier.” Aymonod gives a concrete example: when AI produces meeting notes, it can remove details that appear secondary but may actually matter in the context of the assignment. A good consultant is therefore not only someone who can summarize quickly, but someone who knows what should not be left out.

The judgment paradox
Judgment does not develop automatically. It is built through practice. Writing meeting notes forced junior consultants to listen carefully, select relevant information, synthesize it and structure it. Research tasks required them to understand sources, compare them and challenge them. These activities therefore had two functions: they created production value for the project and learning value for the consultant.
This tension echoes what Bainbridge (1983) described as the “ironies of automation”: the more routine operations are automated, the more humans are needed when automation reaches its limits, even though they may have fewer opportunities to practise the skills required to take control.
Yet AI does not necessarily remove learning; it can also reshape it. Aymonod suggests that some junior consultants may be able to “skip steps” and move earlier towards the relational, commercial and industry-focused dimensions of the job. Similarly, Lee et al. (2025) show that critical thinking does not disappear with AI but shifts towards verifying, integrating and supervising its outputs. Reviewing AI-generated work can therefore become a new form of learning, provided that it is active, structured and supported. Otherwise, review risks becoming little more than passive acceptance of a ready-made answer.

The scarcity of judgment is also a management choice
When a task falls from four hours to five minutes, the time saved can be used in two very different ways: it can be converted into more output, or reinvested in verification, client exposure, expert discussions and reflection.
Aymonod explains why this choice matters. As AI saves time, consultants are often “given more and more to do”; under pressure, the extra thirty minutes that could have been spent reviewing the work may be exactly what disappears. He even recalls having to send back AI-generated work that had not been reviewed to a director, not a junior consultant. The risk is therefore not simply generational. It also depends on how work is organised.
Preserving judgment does not mean banning AI or artificially recreating every traditional junior task. The objective should instead be to preserve the learning mechanisms embedded in those tasks. Two practices could help. First, consultants could be asked to form an initial hypothesis or recommendation before turning to AI. Second, AI review could become a deliberate learning exercise, with juniors expected to identify assumptions, verify sources, spot missing elements and justify their corrections to a senior.
AI may be able to industrialize analysis. The next competitive advantage for consulting firms may be their ability to industrialize the development of judgment as well.

• Aymonod, X. (2026) Interview conducted as part of the Audencia article competition.
• Bainbridge, L. (1983) ‘Ironies of automation’, Automatica, 19(6), pp. 775–779.
• Brynjolfsson, E., Li, D. and Raymond, L.R. (2025) ‘Generative AI at Work’, The Quarterly Journal of Economics, 140(2), pp. 889–942.
• Dell’Acqua, F. et al. (2026) ‘Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of Artificial Intelligence on Knowledge Worker Productivity and Quality’, Organization Science, 37(2), pp. 403–423.
• Lee, H.-P., Sarkar, A., Tankelevitch, L., Drosos, I., Rintel, S., Banks, R. and Wilson, N. (2025) ‘The Impact of Generative AI on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects From a Survey of Knowledge Workers’, Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems, Article 1121, pp. 1–22.

*Quotes from the interview were translated from French by the authors.

Article written by Alexis Launois and Servane Cloix.