The end of the junior consultant — what dies, what is born
Ninth piece in the cognition / doctrine block. After agentic piloting, a subject that imposes itself without our having chosen it: what becomes of the junior consultant, now that AI does, better and faster, just about everything that made up their apprenticeship?
Over these past eighteen months, we have watched an entire stretch of a profession many of us have practiced for a long time fade away without anyone really speaking of it. Not at conferences. Not in the business press. In the hiring figures of master's graduates, in the substance of the conversations between partners at the close of an engagement, in the glances exchanged in a recruitment committee when someone, out of habit, proposes an intern on a pre-hire track. The profession of junior consultant, as it was taught in 2024 and as it was billed, has all but ceased to exist in 2026. And no one knows how to say aloud what that means.
This article does not settle the matter. It tries to set down, in a substance still alive, what has melted away, what holds firm, what is hiding. It walks the thin wire between sociological observation and prescription, because it is on that wire that the debate we never managed to hold is being decided.
The archive of disappearances
We must begin by placing the subject on the shelf where it belongs, which is not the shelf of the unprecedented. What is happening today to the junior consultant happened in the 1980s to the architecture student, and in the 1990s to the secretary. The parallel rule, the ruling pen, India ink on tracing paper, the patient model in foam board: that is what was taught in schools of architecture forty years ago, and it was not merely a technique. It was the root of a judgment. You understood a volume by tracing it; you internalized scale by measuring it. Digital modeling killed that entry gateway in fifteen years. The schools first refused, then adapted, then forgot they had ever resisted. Architects still exist.
The secretary, for her part, typed, took shorthand, filed, archived. Word processing, then email, then the AI agent erased each of these gestures in turn. The profession survived in another form. It became interface, coordination, the living memory of the organization. The function remained, the tooling changed, the root was passed on through the new tools in the structures that knew how to pass it on.
We can draw from this both a consolation and a warning. The consolation is that the consulting profession will not disappear. The warning is that the structures that stalled for ten years before teaching CAD produced a poorly trained generation, one that draws plans without understanding space. The risk, today, is not that consulting dies. It is that it survives by producing AI operators convinced they are consultants.
The vacant perimeter
So as not to remain in the abstract, we must name what has been erased. Reading two hundred pages of a CCTP to extract the dozen that matter. Converting a rough train of thought into presentable slides. Compiling a sector benchmark from the web and the trade press. Turning a ninety-minute meeting transcript into a usable summary. Making a client interview transcript workable. Producing the first iteration of a framing note. Finding comparable cases. Mapping the actors of a topic. That was, until yesterday, the daily lot of a junior: what filled their weeks, justified their day rate, and provided, without anyone ever putting it into words, the terrain through which, after two or three years, they eventually came to recognize sector patterns on their own and to calibrate a judgment.
All of this is now done by an AI agent in a few minutes, at a level equivalent to that of a master's graduate, sometimes higher. What has not disappeared is the quality of a senior who has practiced for fifteen years the gesture they are now piloting. The nuance is not minor. The entry gateway did not melt away because the juniors were mediocre. It melted because the machine reached, on the narrow perimeter we made them occupy, the level we expected of them.
Conversely, what the machine does not do turns out to be the essential thing. Steering an agent mid-engagement, which the previous article already discussed. The fine-grained knowledge of a client, of their internal tensions, of their unspoken assumptions, of what must be heard behind the official wording of a mandate. The ability to ask, in an interview, the question that unlocks things. The ability to catch the silence that signals unease. The ability to follow up without putting someone on the defensive. Physical presence in a strategic meeting. Arbitration in an emotionally tense situation. The building of trust over ten years. All those things we believed to be the summit of the profession, and which become, through the erosion of the base, its condition of entry. The pyramid inverts, gently, and without a sound.
The trapezoid
Hiring thirty juniors a year no longer makes sense when the machine does, at roughly a hundredth of the cost and with a lower variance in quality, what we used to have them do. The first wave has been observable since late 2025: a quiet freeze on master's-level recruitment in several large firms, little publicized but legible in the hiring figures of graduating cohorts. The second is beginning now, through unreplaced attrition. Those who leave in their second or third year are now replaced only one-third of the time, sometimes one-quarter. The pyramid is deforming. It becomes a trapezoid, then will tend, in certain firms, toward the silhouette of a shorter, wider rectangle. The word workshop fits better than firm to describe what is being built: a small number of equipped seniors, rather than a young crowd to be directed.
This deformation kills, in the same movement, the economic model that carried it. Billing forty hours of a junior at two hundred euros, when the buyer knows AI would do it in one hour, becomes indefensible. More precisely: it becomes a priced lie that no one any longer has the decency to buy. The fixed fee per deliverable, per decision, per value imposes itself, and with it a difficulty the firms have not yet resolved, which is to articulate what justifies the price when the effort is no longer visible. Those who did not begin this shift in 2024 are now paying, and will pay more in 2027, the cost of having believed the subject could wait.
A sensor whose existence is discovered as it is lost
There is something, on this subject, that you hear formulated nowhere. Juniors were not merely executors. They were the firm's eyes on the ground. They listened in meetings without really being asked to. They caught, at the coffee machine, the corridor rumor that reveals that the official project is not the real project. They decoded the tone of a business-unit director in a steering committee, the silence of another, the grimace of a third. They brought back to the firm, in a quick briefing between two doors, those observations that never carry the status of a deliverable, that are not billable, that sometimes cannot even be put into words, and that nonetheless, six weeks later, unlock an engagement that all the formal analysis was leaving in the wall.
It was a distributed sociological sensor. Its value was invisible. It is becoming the hidden cost of the end of the junior. The seniors who pilot their agents from a distance recover only part of this information, and even then, the least rich part. AI does not pick up the tension between two departments that you read in people's posture in a meeting. It does not read the silence of a director whose duration signs a political disagreement he will never formulate. It does not grasp the half-overheard conversation in a corridor, which contains the only clue to the project that will appear in the official mandate four months later.
There remains the question, to which no firm director has, to this day, seriously responded: with what do we replace this watch? Sending seniors on site more often means saturating an already strained schedule. Revaluing juniors as sensors and redefining their function around their field presence rather than around their analytical output is an interesting path, but one that requires changing the commercial discourse. Installing permanent staff at the client's premises, in the manner of resident consultants, is costly and structurally difficult. None of the three suffices on its own, and the firms that have not anticipated this gap will discover in eighteen to twenty-four months that they are deciding for their clients without understanding them as finely as before. The slippage will not be spectacular. It will be slow, almost imperceptible, and difficult to reverse.
What the signature now costs
Here is the other angle no one is holding. The junior signed little. They prepared, and the senior validated. The signature that went out of the firm carried work that several pairs of eyes had filtered, and the partner's liability rested, in practice, on this internal audit chain whose value was never calculated because it went without saying.
This chain is disappearing. The senior, in 2026, signs directly the output of an agent they piloted but did not, themselves, write line by line. No junior team to intercept the gross error. No manager to spot the ambiguous wording that could, in a dispute, be turned against the firm. The disputes on the horizon, which we will see rise from 2027 or 2028, will turn principally on three questions: did one verify what the agent asserted, and with what diligence (duty to advise); does a false assertion produced by AI and signed engage the firm in the same way as a validated error by a junior (manifest error); how far can one sign an analysis in a domain where one is not oneself an expert but where the AI appears competent (false expertise)?
The consequence, which we guess at more than we measure it, is that professional liability insurance premiums will rise, and that the senior now carries a personal exposure heavier than before. The commercial discourse that presents AI as the senior's assistant, as what saves them time, must be completed with a less flattering but more exact reading: AI is also a multiplier of their individual exposure. What we save them in juniors, we make them pay for in risk. The balance is not neutral, and it ought to be written into their day rate.
What becomes of the master's degree
Master's programs in strategy and consulting had been calibrated, for thirty years, to train the junior whose disappearance we have just observed. Analysis methods, classic frameworks, case exercises, standardized vocabulary. All of this, any AI agent now restitutes in a few seconds, at a level often higher than what an average student reaches after six months of cramming. The reframing is observable, in the schools that did not wait. What is now taught is the piloting of agents, applied practical epistemology, the political analysis of organizations, the conduct of relationships and interviews, strategic speaking. What remains of Porter and the McKinsey 7S becomes general culture, more ardent than operational.
The schools that will, out of laziness or want of imagination, stack a prompt-engineering course onto an unchanged curriculum will manufacture students poorly positioned for the decade that is beginning. Those that rethink their offering in depth, and that dare to ask their alumni what the profession has really become, will gain a lead whose magnitude they do not yet measure.
The passers-on
There remains, and this is the heart of the matter, the question of how a junior can still, in this landscape, become a senior in the true sense. The calibration of judgment was achieved, in the past, through the accumulation of hours. One recognized, through repeated exposure, the sector pattern; one calibrated, through trial and error on progressively more complex analyses, what would later be called, coyly, experience. If AI absorbs those hours, how is calibration built?
The answer has already been hinted at above, and is not unprecedented. It played out for the architect who learned through the ruling pen, for the secretary who learned through shorthand. The elders become passers-on. Those who did it by hand know the root of the gesture. They know, because they lived it, why one synthesis captures the essential when others miss it, why one question reveals an unspoken thing, why one market analysis holds up under challenge while another collapses at the first follow-up from a demanding buyer. They do not pass on the gesture as such; the tool that carried it is dead. They pass on the root of the gesture, in the new tooling.
Concretely, in a firm that takes this subject seriously, the junior does not produce a synthesis by entering a prompt. They produce it with a piloted agent, under the eye of a senior who shows them why the agent slipped on a given paragraph, how the drift signal should have been spotted, what the client is really seeking behind the official wording of their request. The junior learns the root, differently from their elders. They learn it by piloting, under correction from someone who knows how to recognize the drift because they, themselves, experienced it by hand.
This resembles apprenticeship more than training. It presupposes elders one accepts not to saturate with their own billable files. Juniors one accepts to see progress slowly, in contradiction with the pressure for immediate productivity. Leadership that accepts not to extract one hundred percent of return from senior time in the first eighteen months of a newcomer's career. It is an investment whose return spreads over five to ten years, and whose profitability, like all investments in transmission, is measured only at the moment when one would have wished to have made it. The firms that do not engage in this discipline train operators without roots, who produce defensible deliverables without knowing why, and who remain, under the title of senior that will eventually be given to them, slightly more experienced AI users. The difference from a true senior is not minor. It is exactly the one that separates, today, the architect who masters their CAD because they understood space, from the software operator who produces plans without grasping what they represent.
The summit that rises
If the base loses value, the summit, by mechanical effect, gains it. A senior who pilots five complex engagements in parallel today produces what a team of twenty people did three years ago. Their rarity increases, and so does their personal risk, as we saw. Both play in the same direction. Rates rise in steps, never linearly. The client pays dearly for the senior's decision, not for the delegated execution. An important collateral effect: the figure of the solo consultant, or of the pocket firm of three or five partners without juniors, becomes economically viable under unprecedented conditions. The consulting market will fragment, and certain large firms will discover, in five years, that they have less brand advantage than they imagined against these lighter structures.
For the young consultant who hesitates, here is what can be formulated. The idea that one enters as a junior to become a senior by the usual routes is, in 2026, a trap. The usual route no longer exists in its historical form. Three trajectories seem solid, none of which resembles what one would have advised twenty years ago. Early mastery of agentic piloting, which partially short-circuits the classic entry gateway by making useful, from the first year, a young person capable of steering an agent on complex engagements. Early investment in the relational dimension, in field presence, in the conduct of interviews, in the political reading of organizations, in the slow building of a client relationship. It is slow to bear fruit, but it is what will remain, for a long time, out of the machine's reach. And deep vertical specialization in a sector domain — health or defense or rail or water treatment — to a level of fineness that AI does not reach from its public corpus. Truly fine, situated, up-to-date sector knowledge, acquired through immersion and private conversation with the actors, remains an asset the machine does not reproduce.
To these three trajectories must be added a meta criterion that experience suggests is the most important, without our being able to prove it otherwise. Choose a firm where elders actively pass on the root, and not merely the method. Not the brand, not the entry salary, not the prestige engagement: the quality of the transmission. It does not show in the first year. It shows, decisively, in the fifth. When you discover that you have become a competent operator without judgment, or else a senior pilot with roots, depending on the luck you had at twenty-four.
To the leadership that stalls
For a firm's leadership in 2026, the subject is no longer an AI transformation. It is a structural change in the economic model, one that will render obsolete, in five years, those who have not decided. Redrawing the headcount pyramid over twenty-four months — how many seniors, how many fewer juniors, what geography of expertise, what level of AI investment per senior profile: these arbitrations can no longer wait for the market to stabilize. They must be set down now, even if they are adjusted afterward. Migrating billing toward the fixed fee per deliverable, per decision, per value: a long shift, eighteen to thirty-six months, that demands an overhaul of framework contracts, of rate cards, and of the commercial discourse. The later it starts, the more it costs. Investing, seriously and without false modesty, in inter-generational transmission, by identifying the elders who possess, together, the root and the pedagogy, and by giving them time one accepts not to make profitable. It is the firm's only long-term capital. Integrating, finally, into the risk calculation, the senior's new signatory exposure, renegotiating insurance coverage, documenting internal validation processes, training partners in what they can and cannot sign in the current state of the products. Unglamorous defensive work that will be, in a certain number of disputes to come, the difference between the firm that survives and the one that pays.
The elevator that closes
The last point is the hardest to formulate without falling into the slogan. The junior gateway has been, in France as elsewhere, one of the last social elevators of tertiary capitalism. The door through which profiles that did not come from the grandes écoles, or that came from less privileged backgrounds, could make a place for themselves in consulting, in law, in audit, in investment banking. One entered with an average degree or an atypical training, one imposed oneself through accumulated work, one became a partner after fifteen years. It was unequal, it was hard, it was often unjust, but it was possible. The firms had made, over the past ten or fifteen years, slow but real progress on diversity of origins.
If one removes the gateway and recruits directly at a level of judgment seniority, one recruits, mechanically, on indicators that are historically biased toward the privileged classes. Cultural capital, network, relational ease, political sense in meetings — what Bourdieu called the baggage, and which is transmitted within the family before professional entry. These assets are not built through accumulated hours of work. They are transmitted through origin. If the route through hours disappears, and only the route through baggage remains, the sociological closure of the profession intensifies, and could undo in five years the progress that took fifteen years to build.
No one, in 2026, has formulated this risk publicly. A few HR directors are beginning to speak of it internally, in low voices. The schools that had launched active diversity policies are wondering whether they will have an outlet to offer their atypical students. The end of the junior is not merely an economic matter for firms. It is a political question of the closure of a profession that was, despite its visible defects, one of the last places where a non-linear path could lead to a recognized professional life. The debate will not be posed in these terms in boardrooms; that is not where it is posed. It will be posed in ten years, when we look at the sociological composition of the new cohorts of partners and observe, without being able to undo it, that it has narrowed.
And the subject overflows consulting. It holds for finance, for law, for audit, for investment banking, for engineering consultancy. Every profession with a pyramid of analysts sees its entry gateway contract. All risk, for the same reason, the same narrowing.
Coda
What dies, then, is the junior as a cog. The entry gateway as the sector understood it. Hours sold as a model. The pyramid as a structure. Consulting's capacity to function as an open elevator. A share of distributed field watch that no one named. A certain legal security of the senior, which rested, without saying so, on an internal audit chain that has disappeared with it.
What is born is the senior pilot as the central actor. Seniority as a dearly paid rarity. The profession as an art of relationship and judgment, rather than as an analytical production line. Inter-generational transmission as a distinct strategic asset, one that must be funded and protected. Signatory risk as a new dimension, one that must be integrated into the rate card and into the internal organization.
What remains to be invented is the route to becoming a senior in a world that no longer has a classic junior gateway. It is not unprecedented, as we have seen. It will play out here as it played out for the architect and for the secretary, through elders who pass on the root across the new tools. But it is not automatic. It presupposes firms that accept the investment, schools that reinvent their programs, seniors who accept to pass on, juniors who accept to learn differently, and a little political discipline to prevent the transition from socially re-closing a profession that took thirty years to crack open.
The firms that will have engaged in these things seriously in 2026 will have, in fifteen years, a structural advantage that their stalling competitors will not have seen coming. And they will have contributed to keeping consulting in existence, otherwise, as a profession of thought, and not as a workshop of AI operators. That is not nothing.
TITAN, TenderGraph's cognitive system, applies these best practices — language science, rhetoric, cognitive science — to every bid it works on. And TenderGraph helps teams optimise their pre-sales process along these lines: discover TenderGraph · talk to our team.
Main sources — economics of consulting: Maister, Managing the Professional Service Firm, Free Press, 1993. Greenwood, Suddaby & Hinings, "Theorizing Change," Academy of Management Journal, 2002. Empson, Leading Professionals, Oxford University Press, 2017. — Automation of intellectual work: Brynjolfsson & McAfee, The Second Machine Age, Norton, 2014. Susskind & Susskind, The Future of the Professions, Oxford University Press, 2015. Felten, Raj & Seamans, arXiv 2303.01157, 2023. Brynjolfsson, Li & Raymond, "Generative AI at Work," NBER, 2023. — Acquisition of expertise: Ericsson, Peak, 2016. Klein, Sources of Power, MIT Press, 1998. Schön, The Reflective Practitioner, Basic Books, 1983. — Sociology of professional closure: Bourdieu, La Distinction, Minuit, 1979; La Noblesse d'État, Minuit, 1989. Abbott, The System of Professions, University of Chicago Press, 1988. OPIIEC 2020-2025 surveys, Syntec Conseil barometers. — Legal liability and AI: European AI Act 2024; Lefebvre Dalloz commentaries, La Semaine Juridique, 2025-2026.