Chapter 4 — How Expertise Is Actually Built (plain-language version)
The purpose of this chapter
Before the thesis can explain what AI mediation is changing, it has to explain what was there to be changed. This chapter describes how expertise in European infrastructure advisory is actually formed — at three levels simultaneously: the individual consultant, the firm, and the wider professional domain.
The chapter doesn’t yet make the argument about what AI tools are doing. It just describes the system as it has historically worked, so the subsequent chapters have something specific to analyse.
Level one: how an individual consultant becomes competent
Expertise doesn’t arrive with a degree. It builds through three components that work together.
Formal training. Consultants enter with qualifications — in economics, engineering, urban planning, law, public policy. They continue developing through professional development and formal certification. This provides the analytical vocabulary and disciplinary grounding. Necessary, but not sufficient.
Engagement with consequential work. The core of professional formation. A junior consultant builds contributory expertise — the ability to make genuinely independent analytical contributions, not just apply frameworks — through sustained, difficult engagement with real work: building financial models from scratch, drafting analysis that gets torn apart and rebuilt, attending regulator meetings and learning to read what’s actually being communicated, presenting to clients and learning when to hold a position and when to revise it.
One example: Stockholm, 2018. A young analyst runs a stakeholder mapping workshop for an electricity transmission regulatory engagement. She produces a clean, well-structured map. The partner reviews it and asks: where is Svenska Kraftnät? The transmission operator isn’t directly in scope — but its infrastructure investment programme constrains every strategic option the client is considering. The analyst had mapped the regulatory space. She hadn’t yet learned that the regulatory space sits inside a physical topology with its own logic. She spends the evening redrafting. She has never since produced a stakeholder analysis without asking what the physical network requires. The formation happened in the correction, not in the original output.
Senior judgment witnessed and absorbed. Juniors develop partly through watching senior practitioners exercise professional judgment — in client meetings, in internal discussions, in the moments when senior practice is visible to someone who is still being formed. This knowledge is largely tacit: it can’t always be articulated as a rule. It transmits through proximity, not instruction.
One example: Brussels, 2020. An advisory team is preparing to challenge a regulatory determination. The case is solid. The senior partner — who has worked with Belgian energy regulators since the regulator was established in 1999 — says: not now. He can’t fully explain why. Something about the institutional temperature, a signal in recent speeches, a relationship the client can’t afford to damage. The team thinks he’s overcautious. Eighteen months later, the regulator initiates a sector-wide investigation that directly affects the client. The partner had read something that existed only in decades of accumulated engagement with that specific institution. It wasn’t in any document. It couldn’t have been transferred through a briefing.
These three components together — formal training, consequential work, witnessed judgment — produce what the formation system is designed to produce: contributory expertise. Not just interactional fluency (the ability to speak the language of the domain) but the ability to make genuine independent contributions to it.
A note on history: pre-AI formation wasn’t uniformly excellent. Some formation environments were rich, others thin. The thesis isn’t claiming the past was golden — it’s claiming that AI mediation constitutes a structural alteration to the conditions under which good formation was possible.
Level two: how a consulting firm has its own identity
A consulting firm is not just a collection of individual consultants. It has its own identity — a distinctive way of seeing problems, a characteristic analytical voice, methodological reflexes built up over years or decades of work in specific sectors and with specific clients.
Knowledge management cycles. Researchers Nonaka and Takeuchi described how knowledge moves through organisations. Tacit knowledge held by experienced practitioners gets made explicit in frameworks and playbooks. Those explicit artefacts get absorbed by newer practitioners and become tacit knowledge for them. Repeat across generations. The result is an organisation that knows things no individual holds entirely.
But more importantly: the firm’s knowledge management is identity-producing, not just productivity-supporting. What the firm reproduces through this cycle isn’t just information that helps people do their jobs — it’s the content that makes the firm a particular kind of firm. The specific framings, the sectoral instincts, the vocabulary in which the firm recognises a situation. This is what makes one infrastructure advisory practice meaningfully different from another even when they sell similar services.
One example: a mid-sized London advisory firm has a concept called “the regulatory clock” — an informal framework for reading where any given regulatory process is in its cycle and what that implies for what can and cannot be moved. It’s not in any playbook. It lives in onboarding conversations, in casual annotations, in the way meetings get opened. An analyst from a larger firm joins and spends three months confused by its absence from any document. By month six she’s using it instinctively. By year three she’s teaching it to someone new. Nobody decided to transmit it. It reproduced itself through the practice.
The thesis distinguishes two types of firms with different vulnerabilities:
Large advisory firms have their distinctive identity mainly in institutional architecture — the methodologies, case databases, training curricula, cultural vocabulary. Individual practitioners matter, but the firm’s identity doesn’t depend entirely on any one of them. The risk from AI mediation operates mainly at the institutional level.
Small boutique practices have their distinctive identity concentrated in the accumulated judgment of a small number of senior practitioners. There may be no elaborate methodology to extract. The risk operates more personally — whether the individual’s distinctive analytical voice survives being gradually displaced by generic AI defaults.
Level three: what the domain itself requires
European infrastructure advisory as a domain — not just as a market — requires specific things from its practitioners that distinguish it from generic management consulting.
Technical depth and political judgment, simultaneously. Infrastructure work sits at the intersection of hard analytical competence (financial modelling, regulatory economics, engineering assessment, environmental analysis) and political judgment (reading institutional dynamics, understanding what a regulator is actually worried about, knowing which stakeholders have real weight). Neither component alone is sufficient. The integration of both is what makes the advisory work distinctive.
Contested knowledge as the norm. Infrastructure decisions are genuinely contested. The same project can be analysed through a cost-benefit lens, a climate-transition lens, an equity lens, or a strategic-autonomy lens — and different frameworks produce different conclusions. Advisory work doesn’t exist in a world of clear right answers. It exists in a world where different legitimate frameworks conflict and the adviser has to navigate that honestly.
One example: London, 2019. An advisory firm is commissioned to produce an economic assessment of the Heathrow third runway. The technical case is broadly supportable. But the climate compliance question — whether the expansion is compatible with UK carbon budgets — is contested between the Department for Transport and the Committee on Climate Change. In a parliamentary select committee hearing, the senior partner is asked whether the analysis assumes climate compliance or treats it as a separate policy question. He says it’s a separate policy question — technically correct, given the scope of the commission. But he’s also, in that answer, taking a position about what the advisory role is for: whether advisers are responsible for the normative consequences of their analytical framings or only for the technical correctness of their models. The room knows it. He knows it. He’s thought about it since.
Long time horizons. Paris, 2023. An advisory firm is engaged on procurement governance for the EDF nuclear decommissioning programme — a sixty-year horizon, multiple generations of contractors. A colleague who has advised on airport privatisation thinks the caution about framing is excessive. The senior consultant explains: in nuclear, a wrong framing embeds itself in a procurement framework that governs decisions thirty years from now. The consequence of getting the framing wrong isn’t a bad report. It’s a governance structure that can’t be corrected by anyone who comes after us.
How the three levels constitute each other
The most important claim of this chapter is that the three levels aren’t independent. They produce one another.
The individual is shaped by the firm and the domain. A Foundation-stage consultant formed at one firm in one subdomain becomes a different practitioner from one formed at another firm in another subdomain — even with equivalent formal training and personal ability. The formation is integrated.
The firm is shaped by the individuals it forms and the domain it works in. A firm that has worked extensively on European water utility regulation develops a different analytical character from one that has worked on energy market design, even with formally similar methodologies. The domain shapes the firm’s identity.
The domain is shaped by the firms working within it. European infrastructure advisory as a domain is reproduced across generations by the engagements firms undertake, the knowledge they develop, the institutional relationships they maintain.
One example of all three at once: The Hague, 2017–present. A Dutch advisory practice founded in 1998 has worked almost exclusively in water and environmental infrastructure. Over twenty years it has developed a way of framing regulatory problems that treats the infrastructure asset as a service-delivery system first and a cost-recovery problem second. This isn’t the default framing in regulatory economics. It’s the framing the Dutch water sector’s public-service tradition produces — a tradition the firm absorbed through decades of client relationships. A new partner joins in 2019 with a background in energy regulatory economics. Within eighteen months she’s framing water problems in the service-delivery register. Nobody instructed her. The clients ask questions in that register. The firm’s reputation is built on answering in that register. The domain, the organisation, and twenty years of practitioners being formed through that domain have together produced a way of seeing that recruits her without coercion.
This mutual constitution is important for understanding what AI mediation threatens. A transformation at any one level propagates to the others. If the formation of individual practitioners is structurally altered, the firms that depend on that formation are altered. If the firm’s distinctive way of seeing is eroded, the domain loses some of what constitutes it. The integrated system is not robust to disruption at a single level.
What has been established
The three-scale integrated system — individual formation, firm identity, domain character — is the content that is at stake in the analytical chapters that follow.
Chapter 5 asks what the political economy of AI mediation is doing to this system: who owns the tools, whose cultural patterns they carry, and what happens to European advisory practice as they become embedded in daily work.