Chapter 1 - Introduction

1.1 The puzzle

Lisbon, late 2025. A junior consultant on a railway regulatory engagement produces the risk-framing section of a client deck. Their writing is fluent, the structure is sound, and the regulatory categories are correctly named, but the senior partner reads it and stops at the third paragraph. The section frames the regulatory risk as a generic concession-renewal issue, which follows the textbook, and is wrong. She had spent three meetings identifying the rail regulator’s specific concern. It was not the renewal itself, but the interaction between renewal timing and a pending parliamentary inquiry that the regulator could not name publicly but had signalled obliquely through the questions it kept returning to. The junior had not been in those meetings; nobody had thought to bring them. They had used an LLM to get a draft from the brief. The output was good enough to survive a first read, but not good enough to survive a read by someone who had been in the room when the regulator’s actual concern became visible. The senior partner makes a mental note. She corrects the draft, says nothing, and then spends two weeks thinking about what it means: the junior could not have known what they did not know, the tool could not have known it either, and the quality-control system had nearly missed it. 💡 💡

I have spent a (fair) bit of my career advising on (European) infrastructure projects, enough to recognise when something in our perception of professional work is shifting beneath the surface of ordinary practice. The shift is not really dramatic, as engagements continue, deliverables get produced, recommendations are made and clients receive 💡 them. The advisory profession carries on, and on most days it functions much as it did a couple of years ago. But underneath the visible continuity, something is changing in how the work is being performed, and more importantly, in how the next generation of consultants is being formed.

The puzzle that drives this work is the gap between two facets of how contemporary advisory practice is being talked about.

1. In the first facet (the one that dominates professional discussion, vendor marketing, and the early academic literature on AI in consulting) the use of generative AI tools is presented as banal: a matter of productivity gains, workflow optimisation, the routine integration of new tools into established practice, CV drafting for bids, first go at slides, document summarisation, search and synthesis at scale. The facet is technical and instrumental; it treats AI mediation as an efficiency question to be managed through appropriate task-technology fit, deployment governance, and training in effective tool use.

2. In the second facet 💡 something more substantial is going on. We sense, without always being able to articulate, that the conditions under which professional judgment is formed and exercised are being altered, not just made more efficient. The early-career consultants we are working with are not the same kind of practitioners we were at equivalent stages. The work they produce reads as competent, often as more polished than ours did, but the grasp of the work is sometimes thinner than the polish suggests. The seniors we work for voice concern about formation in terms that productivity framings do not capture. Clients are beginning to procure differently, reflecting their own AI capability development. And underneath all of this, the integrated professional ecology we work within (the firm’s distinctive way of seeing, the European public-interest tradition the work answers to, the relationship between senior judgment and junior formation) feels like it is shifting beyond what the productivity facet can describe.

This work takes the second facet seriously 💡 . It argues that the gap between the banal current use of AI and the structural transformation practitioners sense beneath it is signal rather than noise, and that taking the signal seriously requires a framework the existing literature on AI in consulting does not yet provide.

The framework is philosophical. Its first anchor is Bernard Stiegler’s account of proletarianisation as the selective displacement of savoir-faire into technical systems 💡 that the worker, the firm, and the professional culture do not own 1 2 . Its second is Hans Jonas’s imperative of responsibility 3 , which grounds the answerability of infrastructure advisory to the publics that bear the consequences of decisions. Around these sit a few supplementary supports (MacIntyre on the goods internal to a practice 4 , Polanyi 5 and Collins 6 on tacit and contributory expertise, Dewey on publics formed through shared consequences 7 ) that contribute at specific points in the argument.

The framework lets us describe what we have been sensing in terms that are neither nostalgic nor technologically determinist. I am not arguing that consulting was better before AI tools existed 💡 , or that AI tools should be refused 💡 , but more interestingly that the conditions under which professional advisory work has historically reproduced its expertise 💡 are being structurally altered by new technologies and ways of working 💡 , under specific political-economic conditions.

This work develops and tests the argument in European infrastructure advisory, where the formation and legitimacy stakes are easiest to see 💡 .

1.2 The title and what each component does

The title of this work is The Proletarianisation of Advisory: Judgment, Sovereignty, and Responsibility in European Infrastructure Consulting under AI Mediation. Each component does work in the argument that follows.

The Proletarianisation of Advisory names the conceptual frame. Proletarianisation is taken from Stiegler, who reworks the term from its Marxist origins to describe the loss of savoir-faire (practical know-how, but also know-how-to-live and ultimately know-how-to-think) as it is progressively captured into technical systems the worker does not own 1 2 , and where the firm may lose control to systems providers. Proletarianisation here is selective: it does not abolish the worker but displaces the expertise that constituted their professional identity. This, may we argue, is what is happening to advisory work under generic AI mediation, and it operates on professional formation regardless of whether headcount changes.

Judgment, Sovereignty, and Responsibility names the three main concerns this work develops. Judgment is the contributory expertise (in Collins’s sense 6 ) that distinguishes the senior practitioner from the model: judgment formed through sustained engagement with consequence, exercised in situated contexts, and irreducible to the patterns the model has absorbed from its training distribution. Sovereignty is the dual-scale concern Chapter 5 develops: the distinctiveness of the European public-interest tradition (the national-cultural scale) and the firm’s distinctive way of seeing (the organisational scale), both at risk of erosion under generic AI mediation that carries the discursive patterns of its training distribution. Responsibility is the Jonasian framing 3 of what the work owes to the publics that bear the consequences of infrastructure decisions, including the future generations whose conditions of life are being shaped now.

European Infrastructure Consulting names the primary research scope 💡 . The claim this work advances is general but the domain in which it is developed and tested is specific. It covers advisory work across the family of European infrastructure subdomains (transport, energy, water, defence, environmental, social, built-environment, and cities) that share a public-interest inheritance distinguishing them from generic management consultancy. We can treat it as one family bound by the European public-purpose tradition, while recognising throughout that subdomains themselves vary in regulatory weight, time horizon, and third-box exposure.

Under AI Mediation puts a name on the mechanism. The work is deliberately tool-agnostic: it names no products and does not anchor its argument to the state of any particular model. Its object is the class of socio-technical mediators (generic LLMs operating as inscriptions of collective discourse, recirculated under specific conditions of ownership) 💡 and what their absorption into advisory practice does to the integrated three-scale system through which expertise has historically been materialised. The tool-agnostic stance is a methodological commitment: the technologies we are talking about are changing on a faster cadence than paper-writing can track, and the argument needs to operate at a level of abstraction that survives the churn of specific tools 💡 .

The four components together describe our frame: a philosophical argument about how professional advisory work is being transformed under the mediation of a class of socio-technical systems, with European infrastructure advisory as the primary research site, with consequences for judgment, sovereignty, and responsibility that practitioners and institutions have not yet adequately engaged.

1.3 The central claim and why it matters

This work advances a central claim at two levels. At the general level:

Generic AI mediation, under current conditions, proletarianises the situated, answerable judgment on which the legitimacy of professional advisory work rests.

We try and develop, and stress-test, this claim in European infrastructure advisory, chosen as the primary research site because its public-interest obligations, regulatory architecture, and long time horizons make the formation and legitimacy stakes most legible, and because it is the domain where I have practitioner-level access. In European infrastructure advisory, the claim takes its most demanding form:

Generic AI mediation, under current conditions, proletarianises the situated, answerable judgment on which the legitimacy of European infrastructure advisory rests, and on which its obligations to publics and futures depend.

The “hardest case” logic applies: if the proletarianisation mechanism operates where AI substitution should be least attractive (in high-stakes, long-horizon, publicly accountable advisory work), finding it there strengthens the case for its generality, though it does not prove it.

The argument runs in two registers throughout. In the general register, it makes claims about the proletarianisation mechanism in advisory work broadly: claims about formation pipelines, sovereignty, and legitimacy that apply wherever advisory practitioners are formed and their judgment is at stake. In the specific register, it develops, illustrates, and tests those claims within European infrastructure advisory. Readers working in other advisory contexts 💡 should read the specific-register material as an instance of the general claim, not as its boundary.

Each component of the specific-register claim earns its place. Generic AI mediation names the mechanism, deliberately bracketing tool specifics. Under current conditions signals that the claim depends on the political-economic conditions of model production and ownership the work develops in Chapter 5; the claim is not technologically determinist, and Chapter 8’s scenarios show that alternative conditions would produce different trajectories. Proletarianising specifies the mechanism in Stiegler’s sense: selective displacement of expertise rather than wholesale substitution of practitioners. Situated, answerable judgment names contributory expertise: judgment formed through sustained engagement with consequence, reproduced through the integrated three-scale system Chapter 4 describes, and answerable to the publics that bear the consequences of infrastructure decisions. Legitimacy signals that the stake is the ground on which advisory recommendations are received as warranted, rather than productivity or efficiency. Obligations to publics and futures names the third box and the Jonasian responsibility I treat as constitutive of legitimate infrastructure advisory.

Why does the claim matter? Three reasons, briefly.

First, because it identifies what is at stake in contemporary advisory transformation in terms the productivity framings cannot (and don’t) capture. The existing literature on AI in management consultancy 💡 addresses tasks, workflows, and organisational governance. The framework that literature deploys can see efficiency gains and skill-retention concerns, but it cannot see what happens to professional formation, to firm-level distinctiveness, or to the answerability of advisory work to publics that bear consequences.

Second, because the European public-interest tradition the work answers to is itself at stake. Infrastructure decisions bind futures materially. A grid interconnection, a water tariff reform, a tram extension commissioned today: each shapes how people in Europe will live across decades, in some cases across generations. The advisory profession that supports these decisions has historically operated within institutional architectures that recognise the public-interest character of the work. Whether those architectures stay answerable to the publics they serve depends on whether the profession working within them keeps materialising the expertise on which that answerability has rested. This, let’s argue, is a question AI mediation has begun to pose, and engaging it requires an institutional response that the productivity framings do not generate.

Third, because the argument yields differentiated recommendations for the four constituencies whose decisions shape the trajectory: individual consultants, advisory firms, infrastructure clients, and European policymakers acting on behalf of publics. Each constituency faces a version of the question this work names; each can engage it through the institutional means available to it. The recommendations developed in Chapter 9 answer to four of the seven cross-scenario findings from Chapter 8 (formation, sovereignty, the third box, legitimacy narration) and are robust across the trajectory variation the scenarios capture. The analysis is also practical: it says what each constituency can do, and on what grounds. 💡

1.4 Six “original” contributions

The work sits at the intersection of several active research clusters, but is not fully anticipated 💡 by any of them. The literature on AI in management consultancy is growing rapidly, with the Tuczek et al. (2026) Task-GenAI Fit framework 8 as the most directly comparable contemporary contribution and the broader management and information-systems literature providing additional empirical and conceptual context. The literature on AI sovereignty and cultural bias is well-developed and has produced substantial work on the political economy of AI systems. The literature on AI in adjacent professions (particularly law and software development as we’ll cover them) 💡 has produced empirical evidence and theoretical articulation 9 10 that this work draws on directly. The philosophical literature on technics, individuation, responsibility, and tacit knowledge grounds the framework’s analytical apparatus.

This work tries to bring these clusters together to address a specific question through a philosophical framework suited to the domain, which the combined literature has not done 💡 . Six contributions follow from this combination, each noted here and developed across the analytical chapters.

First, this work applies Stiegler’s proletarianisation thesis 1 2 to management consultancy 💡 . Stiegler’s framework has been applied to digital labour, to platform work, to cultural and educational contexts, and to critiques of contemporary capitalism 11 12 , but its application to the advisory professions specifically has not been undertaken 💡 . The application requires developing the three-scale individuation argument 13 (operating at the levels of the individual consultant, the firm, and the national-cultural tradition) that gives the framework conceptual unity across the analytical chapters. This is a theoretical contribution to both Stiegler scholarship and consulting studies.

Second, this work extends Jonas’s imperative of responsibility 3 to AI-mediated advisory accountability in infrastructure contexts. Jonas’s framework has been deployed in environmental ethics, bioethics, and broad technology ethics, and has been cited in AI ethics literature for its precautionary structure, but it has not been applied to advisory work in infrastructure, where the long-horizon, public-interest, future-binding character of the decisions makes it apt. The application grounds the third-box-thinking argument (what advisory work owes to the publics and future generations who bear its consequences), as well as the sovereignty claims.

Third, this work maps the Chartered Management Consultant (ChMC) competency framework 14 against AI substitutability across the three stages of professional progression. Analogous mappings have been undertaken in legal and medical contexts 9 , where the corresponding competency frameworks (bar admission, clinical training accreditation) have been used as analytical scaffolds for studying AI mediation. The ChMC framework, launched in 2021, has not yet been used in this analytical mode, and this work provides what is, to my knowledge, the first such mapping 💡 . The mapping gives the analysis a stage-by-stage diagnostic that Tuczek et al.’s framework does not provide.

Fourth, this work theorises pipeline rupture as a cross-professional philosophical mechanism, demonstrating through the comparative evidence in Chapter 6 that the mechanism already operates in software and law, and that it operates on professional formation independently of headcount changes. It is the most robust of the six because it survives the AI-washing critique that complicates the labour-market data. The cross-professional empirical pattern (selective displacement, pipeline rupture regardless of headcount, uneven upward value flow, explicit narration of professional legitimacy) is consistent enough across the comparative cases to support the theoretical framing, and the framing itself has not been articulated in this form in the existing literature.

Fifth, this work develops the dual-scale framing of sovereignty as identity-formation under technical mediation, connecting the national-cultural and organisational scales through a single mechanism (dis-individuation under generic mediation). The existing literature on AI sovereignty operates predominantly at the procedural level 💡 , and where it engages sovereignty concerns, it does so primarily at the national or supranational scale. It has not made the move that joins the two scales.

Sixth, this work constructs and analyses four scenarios 💡 for European infrastructure advisory as philosophical thought-experiments rather than predictions, identifying the signals and trends that hold across all four. Scenario methodology has been used in consulting studies and futures research, but its application as philosophical thought-experimentation in the Williamsonian sense 15 , suited to the concerns of European infrastructure advisory, has not been done. The cross-scenario findings 💡 will help ground an institutional response that does not depend on which trajectory unfolds.

These are the six claims to originality. The first (Stiegler applied to consulting) and fourth (pipeline rupture as cross-professional mechanism) are general contributions whose scope could reach beyond European infrastructure advisory. The second (Jonas applied to infrastructure accountability) and third (ChMC mapping) are developed in the European infrastructure context, with implications available to adjacent advisory contexts through analogous frameworks. The fifth (dual-scale sovereignty) and sixth (scenarios as thought-experiments) are contributions whose analytical architecture is general (the sovereignty mechanism operates in any non-Anglo-American advisory context, and the scenario methodology applies to professional advisory broadly), developed in their most specific form in the European infrastructure research site. None of them is the final word on the questions it engages; each opens further work, and Chapter 9 sets out the limits. Taken together, and working through the one framework, they amount to a philosophical contribution to the conversation about AI in professional advisory (and in European infrastructure advisory specifically) that the literature has not yet made.

1.5 Methodology and positionality

This work is reflexive practitioner research. Its methodology has to be declared explicitly because the work is conducted from inside the profession it examines, and the epistemic claims it makes depend on what that position permits and what it constrains.

I write as a working consultant in the European infrastructure space. The brain-itch that drives this work is one I have experienced 💡 , not observed from outside. The lived material the analytical chapters draw on comes from my own practice, exchanges with colleagues, partners and competitors, observation of colleagues across firms and sectors, and the broader professional discourse I am part of. We will try to be explicit about where the analytical chapters draw on this material and where on the published empirical and theoretical literature.

The methodological position has both an epistemic advantage and an epistemic burden. The advantage is access to material that external observers cannot reach: the fabric of bids, the practical weight of deliverables, the political feel of client relationships, the quiet ways in which judgment is formed or foreclosed in daily practice. The burden is an obvious one: vested interest, motivated reasoning, the temptation to mistake one’s own situation for the situation of the field. Both have to be named, and the methodology has to be designed to respect the advantage while disciplining the burden.

We will draw on the reflexive practitioner tradition initiated by Donald Schön 16 , who argued that competent practitioners do not merely apply theory to practice but 💡 reflect in action, generating and revising knowledge through engagement with the situations they confront. Schön’s framework remains foundational, but it has been criticised, particularly by phenomenologists who argue that it is too cognitivist, treating professional knowledge as a matter of explicit reflection rather than embodied engagement. We can still take from Schön the basic move 💡 while taking the phenomenological critique seriously enough to be cautious about what reflection alone can yield. The deeper engagement with Merleau-Ponty’s 17 account of perception and embodied skill, with Dreyfus’s 18 account of skill acquisition, and with the philosophical grounding of reflexive practice in phenomenological vocabulary is deferred to residual work 💡 .

The methodology is also tool-agnostic, and this commitment needs a bit of justifying. Work anchored to specific products or capabilities risks dating itself before it can be properly read. Work that focuses on the technical artefact misidentifies the object of inquiry: the technology is the means by which the transformation operates, not the thing transformed. Professional judgment, formation, and identity are being transformed, and these can be examined at a level of abstraction that outlasts 💡 any particular tool.

Let’s try and commit to discussing the class of socio-technical mediators rather than any particular instance or tool, and not to catalogue prompt libraries, agent frameworks, or model releases. Where specific products matter for the argument, we could cite them as evidence of practice, rather than as the object of analysis.

The methodology of course has its limits, which the conclusion sets out in detail 💡 . This is not a systematic empirical study: we have no interview programme, no survey of consulting practitioners, no longitudinal study of AI adoption in firms 💡 . The central claims are conceptual and normative, and should be read as such; their empirical validation in the European infrastructure context is one of the residual questions Chapter 9 identifies. This framework has been sufficient for the analytical work done here, and it is not the full philosophically complete framework the conclusion will point toward.

1.6 The structure of this work

Nine chapters, each building on what the previous ones have explored, the first being what you are reading now.

Chapter 2 situates the thinking. We will anchor the definition of consulting in the ChMC framework 14 : four competencies (ethics and professional standards, leadership and management, client operating environment, personal and professional development) traversed by three stages of progression (Foundation, Applied, Chartered). We will characterise European infrastructure advisory as a domain shaped by a public-interest inheritance with three components: a tradition of public-purpose engineering, a regulatory architecture for utilities (with energy and water as the most highly regulated subdomains, and nuclear as a particular case of layered governance), and an intensifying EU governance frame around climate transition and strategic autonomy. We introduce the third box thinking 💡 as the constituency to which advisers answer, grounded in Dewey’s account of publics 7 and Jonas’s imperative of responsibility 3 . The chapter also sets out the methodological position of reflexive practitioner research, deepened in Chapter 3.

Chapter 3 assembles our philosophical framework. It builds its spine from two main anchors: Stiegler 1 2 on the proletarianisation of savoir-faire under technical capture and on individuation 13 as the unifying concept that works across three scales (individual, firm, professional culture), and Jonas 3 on the imperative of responsibility toward futures. Supplementary anchors (MacIntyre 4 on internal goods, Dewey 7 on publics formed by shared consequences, Collins 6 and Polanyi 5 on tacit and contributory expertise) are introduced for specific arguments. A dedicated section addresses the distinction between tacit-as-constraint (the reading Tuczek et al.’s framework adopts) and tacit-as-formation (our Stiegler-mediated claim that tacit knowledge is constitutive of professional becoming, such that displacing the tasks through which it is formed is a different and more serious matter than absorbing the output of those tasks). The chapter establishes the working ontology of AI tools as a socio-technical inscription of collective discourse, recirculated under specific conditions of ownership 💡 .

Chapter 4 describes the materialisation of expertise in European infrastructure advisory. The individual scale: how consultants are formed through sustained engagement with consequence, how tacit pattern-recognition cohabits with formal training, how contributory expertise is developed across a career that progresses through the ChMC stages. The organisational scale: how consulting firms operate as knowledge-management collectives, how the SECI cycle 19 20 of socialisation, externalisation, combination, and internalisation 💡 reproduces the firm’s distinctive way of seeing across generations, and how organisational knowledge management is constitutively identity-producing rather than merely productivity-supporting. The domain scale: variation across the family of European infrastructure subdomains in regulatory weight, time horizon, stakeholder shape, and decision character; the unifying public-interest texture; the dual character of technical depth and political judgment; and the role of contested knowledge. The chapter argues that these three scales are mutually constitutive: the individual is individuated by the firm, the firm by the domain, the domain by the firms that work within it.

Chapter 5 develops the political-economic dimension of this work in two parallel arguments. The first, sovereignty, works across two scales: at the national-cultural scale, generic AI mediation introduces a gradient that tilts the texture of advisory work toward the dominant registers of the model’s training distribution and away from the registers distinctive to the European public-interest tradition. At the organisational scale, the same mediation tilts firm-level outputs toward common defaults and erodes the firm’s distinctive way of seeing. The two scales are connected by the same Stieglerian mechanism (dis-individuation under generic technical mediation) at different levels of the integrated system. The second argument, extraction, examines the risk that the firm’s externalised knowledge flows outward into the training materials of the models the firm itself is using, producing the rigorous form of the dark-mirror concern: the SECI cycle’s externalisation step running outward, into systems the firm does not own. The defence reference case provides an institutional anchor, as European policymakers already treat AI sovereignty as a strategic concern in defence, and the logic of that recognition extends to other infrastructure subdomains.

Chapter 6 draws on comparative evidence from two adjacent professions where AI mediation has been operating at scale for longer. The software case provides substantial empirical evidence of selective displacement at the level of individual practice and pipeline disruption at the level of formation, supplemented by labour-market data that requires careful handling because of the AI-washing complication. The law case provides theoretical articulation of the formation problem in Faulconbridge’s 9 mediated-evolution framing, empirical evidence of similar patterns in the legal profession, and institutional-response evidence showing how a regulated profession is beginning to develop explicit governance around AI mediation. The AI-washing critique is met directly, to show that the argument addresses formation whatever happens to headcount. The Susskinds’ Future of the Professions 10 serves as foil: this work offers a formationist account where theirs is substitutionist. Four findings 💡 are synthesised at the end of the chapter and carried forward into Chapter 7.

Chapter 7 is our analytical core. It traces the proletarianisation mechanism through the three stages of the ChMC framework 14 , deploying the TGAIF overlay 8 , the findings from Chapter 6, and the three vectors of transformation (working backwards from the commoditisation of outputs, extraction of inputs, apprenticeship rupture) as analytical lenses applied within each stage. At Foundation, the apprenticeship rupture hits hardest, because AI now mediates the very tasks through which juniors have historically been made into the next generation of consultants. At Applied, the deliverable-as-performance moat erodes as commoditisation absorbs the work that has demonstrated the firm’s distinctive value at this tier, and the supervisory role through which Applied consultants helped form Foundation consultants is altered. At Chartered, disintermediation operates immediately as clients build in-house AI capability that compresses engagement scope, and relational erosion follows as the long-term consequence of pipeline disruption at Foundation stage propagating forward into the senior practitioners of two decades from now. The third-box-thinking runs through the analysis as the constituency to which the argument about what advisers owe publics and future generations always returns. The chapter also notes that the trajectory described is not deterministic; resistant practice exists at each stage, in firms, partners, and practice areas that recognise the formation question and choose differently 💡 .

Chapter 8 takes the cross-stage pattern into futures and futures-gazing. There we construct four scenarios for European infrastructure advisory as philosophical thought-experiments 💡 , derived from two structural axes that emerge from the framework: where expertise lives (in consultants or in models), and under whose sovereignty AI models operate (extractive Anglo-American commercial or alternative) 💡 . The four scenarios 💡 (Commodity Advisory, The Digital Twin Economy, Artisanal Advisory, and Disintermediated Clients) are each read for their consequences across the three ChMC stages, the third box, the two scales of sovereignty, and the extraction dynamic. The cross-scenario synthesis identifies seven findings that hold regardless of trajectory: the distinction between contributory and interactional expertise determines professional value, the formation question becomes inescapable, sovereignty pressure operates across multiple scales, the third-box question persists, explicit narration of professional legitimacy is required, extraction operates structurally, and the Applied stage has no safe trajectory. These findings provide robust ground for the recommendations in Chapter 9.

Chapter 9 is .. the conclusion. It restates the central claim, names the six contributions this work has delivered, and develops differentiated recommendations for the four constituencies this work has engaged throughout: individual consultants 💡 ; advisory firms 💡 ; infrastructure clients 💡 ; and European policymakers 💡 . This last chapter acknowledges its main limits (three: the single-domain character of the analysis; the reflexive and conceptual, not systematically empirical, methodology; and the deliberate lightness of the philosophical apparatus) and identifies a few residual questions that further work could 💡 take up.