There is a conversation that happens inside almost every large consultancy engagement, usually around week six. The partner who sold the work has moved on to the next pitch. The senior manager who ran the discovery phase has been redeployed to a higher-margin account. What remains is a capable but inexperienced team working from a framework built for a different client in a different sector, trying to reverse-engineer context they were never given. For most organisations, this is an inconvenience. For regulated organisations navigating AI governance, it is a material risk.
This article is not an attack on large consultancies. Many of them employ brilliant people and deliver genuine value. It is, however, an examination of the structural incentives that shape what those organisations can and cannot do — and an honest account of why those structural tensions matter most precisely where the stakes are highest.
The Structural Incentives That Shape Large Consultancy Advice
Large consultancies are not charities. They are professional services firms with shareholders, utilisation targets, and growth mandates. Understanding this is not cynicism; it is the necessary starting point for any honest conversation about what kind of advice they are structurally capable of delivering.
The economics of a large consultancy engagement are driven by headcount and duration. Revenue per engagement is a function of the number of people billed multiplied by the number of weeks they are billed. This creates a structural incentive to grow scope, extend timelines, and introduce workstreams that keep the team on-site. Individually, consultants may be motivated by the best of intentions. Institutionally, the firm is optimised for expansion.
Frameworks accelerate this model. A templated governance framework can be adapted and deployed across dozens of clients with minimal senior input once it has been built. It generates margin efficiently. The problem is that a framework built to travel is, by definition, a framework built to fit the average case — not the specific regulatory environment, risk posture, or organisational culture of any particular client. When a regulated organisation receives an AI governance framework, it is frequently receiving a product that has been lightly customised rather than genuinely designed.
There is also the question of risk calibration. Large consultancies carry reputational risk at scale. Advice that is bold, specific, or that contradicts a client's existing direction is advice that can generate conflict, scope disputes, or disengagement. The institutional incentive, therefore, is to calibrate advice to a level of specificity that protects the consultancy while appearing comprehensive. Recommendations are often written to demonstrate thoroughness rather than to enable decision-making.
None of this is malicious. It is structural. And in most commercial contexts, clients can absorb the inefficiency. In regulated environments, they often cannot.
Why Regulated Environments Amplify Every Consultancy Weakness
Regulated organisations — financial services firms, healthcare providers, insurers, energy companies, public sector bodies — operate under a different set of constraints than their unregulated counterparts. AI governance failures in these environments do not produce merely operational disruption. They produce regulatory enforcement action, board-level accountability, reputational damage that is difficult to recover from, and in some cases, direct legal liability for senior individuals.
This changes the nature of what good advisory support looks like. A regulated organisation does not need a framework document. It needs advice that has been stress-tested against its specific regulatory obligations, its existing risk and compliance architecture, and the actual AI systems it is deploying or considering. It needs someone who understands the difference between what the EU AI Act requires in principle and what it requires for a specific use case in a specific sector. It needs continuity of relationship so that the adviser understands not just the current question but the history of decisions that led to it.
Large consultancies struggle to deliver this for structural reasons. The rotating team model means that context is constantly being rebuilt. Each handover introduces the risk that nuance is lost, that a decision made three months ago is not understood by the person now advising on its implications, and that the organisation ends up re-explaining itself repeatedly to people who are billing for the time it takes to catch up.
The framework problem is equally acute. Regulated environments require governance that is mapped to specific regulatory texts, that accounts for the organisation's particular regulatory relationships, and that can be defended to a regulator if challenged. A templated framework is rarely defensible at that level of specificity. Regulators are not impressed by comprehensive-looking documents. They are looking for evidence that the organisation genuinely understands its obligations and has built governance that reflects that understanding.
Finally, risk calibration becomes genuinely dangerous in regulated environments. Advice that is hedged to protect the consultancy may leave the organisation without a clear position on a question that requires one. When a regulator asks how an organisation assessed the risks of a particular AI system, the answer cannot be a reference to a framework document that lists considerations without reaching conclusions. Someone needs to have made a judgment. In a large consultancy engagement, it is often unclear who made that judgment and on what basis.
The Founder-Led Model: Accountability That Cannot Be Delegated
A founder-led AI advisory practice operates on a fundamentally different model. The person who scopes the engagement is the person who does the work. There is no handover to a junior team. There is no account manager mediating between the client and the practitioner. There is no institutional hierarchy that the advice has to survive before it reaches the client.
This creates a form of accountability that large consultancies structurally cannot replicate. When the founder of Navitec AI advises a regulated organisation on its AI governance framework, that advice carries the founder's name and professional reputation. There is no institutional buffer. If the advice is wrong, or incomplete, or calibrated to avoid a difficult conversation, the consequences fall directly on the person who gave it. This is not a rhetorical point about character. It is a structural feature that changes the nature of the advice.
Founder-led practices are also free from the utilisation incentives that shape large consultancy behaviour. There is no partner breathing down the neck of the engagement team asking why the scope has not expanded. There is no institutional pressure to add workstreams to justify continued presence. The engagement is structured around what the client needs, not around what generates the most billable hours. When the work is done, it is done. When additional work is needed, it is identified honestly rather than manufactured.
This also means that difficult conversations happen differently. A founder-led adviser can tell a client that its current approach to AI governance is insufficient, that its proposed AI use case carries regulatory risks it has not adequately assessed, or that the framework it received from its previous consultancy does not meet the standard it believes it does. These are conversations that large consultancies often find difficult to have because they risk the relationship, the renewal, and the upsell. A founder-led practice has a different relationship with those risks, and therefore a different capacity to have those conversations honestly.
Continuity and Independence as Genuine Governance Advantages
In regulated environments, continuity of advisory relationship is not a luxury. It is a governance asset. A regulatory examination, an internal audit, or a board-level review of AI governance may revisit decisions made months or years earlier. The organisation needs an adviser who was present for those decisions, who understands the reasoning behind them, and who can articulate that reasoning credibly and consistently.
Large consultancies cannot reliably deliver this. Personnel move, teams rotate, and institutional memory is stored in documents that were written to be comprehensive rather than to preserve the thinking that produced them. The regulated organisation that has worked with the same large consultancy for three years may find, when it needs that continuity most, that none of the people currently on the engagement were involved in the foundational work.
A founder-led practice provides genuine continuity because there is genuinely only one person to whom continuity can attach. The founder knows what was recommended and why. The founder knows which risks were assessed and which were deliberately accepted. The founder knows the regulatory context that shaped a particular decision and can explain it to a regulator, an auditor, or a board with the same clarity as the original advice.
Independence matters for similar reasons. A senior AI advisory practice that is not financially dependent on a single client and that has no institutional upsell agenda is structurally freer to give independent advice. It is not trying to protect a broader relationship. It is not constrained by what the firm's risk committee will approve. It is not managing the commercial sensitivities of a multi-service relationship where AI advisory is one line on a larger invoice. The advice is independent because the practice is structured to be independent.
For regulated organisations, this independence has concrete value. It means that the governance advice they receive has not been filtered through an institutional lens. It means that the risk assessments they receive are genuine risk assessments rather than risk assessments calibrated to a particular commercial outcome. And it means that when they present their AI governance approach to a regulator, they can describe the process by which it was developed with confidence.
What a Senior AI Advisory Practice Delivers That Scale Cannot
Scale is often presented as a virtue in professional services. Larger firms can draw on broader expertise, deeper benches, and more extensive methodologies. In some contexts, this is genuinely valuable. In senior AI advisory for regulated organisations, it is frequently a distraction from what actually matters.
What regulated organisations need from an AI advisory practice is not breadth. It is depth, specificity, and judgment. They need an adviser who has worked with regulators, who understands how regulatory examinations are conducted, and who can translate abstract governance principles into specific, defensible practice. They need an adviser who can look at a proposed AI system and identify not just the generic risks but the risks that are specific to this system in this regulatory environment with this risk architecture.
A senior AI advisory practice delivers this through the direct application of senior expertise to every engagement. There is no pyramid of resource allocation in which a partner provides direction and a junior analyst does the work. The senior practitioner is the resource. Every document, every recommendation, every risk assessment, every piece of advice has been produced by the person with the expertise and accountability to stand behind it.
This also means that the practice can engage with genuine complexity rather than resolving it into a framework. Regulated AI governance is genuinely complex. The EU AI Act, the FCA's expectations around algorithmic systems and AI, ICO guidance on automated decision-making, sector-specific requirements — these do not resolve into a single framework. They require judgment about how different obligations interact, where the greatest risks lie, and how an organisation should sequence its governance investments. That judgment is what a senior advisory practice sells. It is not something that can be templated.
For organisations at earlier stages of AI maturity, a senior advisory practice also provides a form of education that is genuinely useful rather than performative. Understanding why certain governance measures are required — not just that they are required — is what enables an organisation to build governance capability internally rather than remaining perpetually dependent on external support.
Choosing the Right Advisory Model Before Governance Failures Force the Decision
The conversation about advisory model choice rarely happens proactively. It tends to happen after something has gone wrong: a regulatory enquiry that reveals gaps in AI governance documentation, an audit finding that the organisation cannot explain the basis of a key AI governance decision, an incident involving an AI system that triggers both regulatory and reputational consequences. By that point, the choice of advisory model is no longer strategic. It is reactive, expensive, and constrained by the need to manage an active problem rather than prevent one.
Regulated organisations that are serious about AI governance should make this choice before they are forced to. The question is not simply which advisory model is better in the abstract. The question is which advisory model is structurally capable of delivering what the organisation needs given its specific regulatory environment, its current AI maturity, and the governance challenges it is likely to face in the next two to three years.
For many regulated organisations, particularly those in the early to middle stages of AI maturity, the answer is a senior AI advisory practice. Not because large consultancies are incompetent, but because the structural features of a founder-led practice — accountability, continuity, independence, genuine seniority at every stage of engagement — are more closely aligned with what regulated AI governance actually requires.
The cost of getting this wrong is not measured in advisory fees. It is measured in regulatory findings, board-level consequences, and the reputational damage that follows an AI governance failure that a more independent, more accountable adviser might have prevented. That is the calculus that regulated organisations should be making when they decide who advises them on AI governance.
Navitec AI works exclusively with regulated organisations across all stages of AI maturity. Every engagement is led and delivered by the same senior practitioner, with no delegation to junior resource and no agenda beyond the client's governance outcomes. If you are evaluating your AI advisory approach, we would welcome a direct conversation.