How consultancies can scale with AI without leaking client trust
The consultancy's AI dilemma
Clients want faster answers at lower fees. Consultancies want to deliver that without hollowing out their margins or their people. Generative AI looks like the obvious bridge. Research that took a week now takes an afternoon. First drafts write themselves.
The catch is that the same technology also undermines how the work is priced. And it puts the one thing a consultancy cannot afford to lose at risk: client trust.
Why AI is breaking the billable hour
For decades, consulting has run on a simple equation: time spent multiplied by rate. AI breaks that equation, and there is hard evidence for it. In a field experiment with Boston Consulting Group, Harvard Business School researchers gave 758 consultants realistic consulting tasks. Those with access to GPT-4 completed 12% more tasks and finished them 25% faster, with work rated of higher quality. The same study found something that matters later in this piece: on a task outside the model's competence, consultants using AI were 19% less likely to get the answer right.
When an analysis that took a team three days can be produced in three hours, hours stop being a defensible unit of value. Clients know it. They ask why the invoice looks the same when the work took a fraction of the time, and some already push for fixed prices, outcome based fees, or a discount on anything the AI produced. The large firms are already moving. McKinsey disclosed in late 2025 that about a quarter of its global fees now come from outcome based pricing rather than hours. The billable hour will not vanish overnight. Its dominance is ending, though.
So if hours are no longer the product, what is? The answer the client cares about: the insight, and how fast it arrives. Firms will compete less on how many people they can staff and more on how good the answers are that they put in front of the client. And underneath that sits a question most firms have not yet asked themselves: how far can the client trust the AI behind those answers?
That is where the race will be decided, and the firm that adopts AI first will not necessarily win it. The firm whose AI is most reliable will. Reliable here means answers grounded in the firm's own expertise and traceable to their sources, running on infrastructure that keeps each client's data separate. A fast tool that leaks data, or cannot pass a client's security review, is a liability dressed up as an advantage.
The trust paradox in AI for consulting
A consultancy's most valuable asset is the confidential information clients hand over. Financials, strategy, M&A plans, internal organisation data, sometimes personal data on employees and customers. The slide template and the methodology come a distant second.
Shadow AI is already on the team
Consultants under deadline pressure reach for whatever works. They paste a client's cost breakdown into a consumer chatbot to summarise it, or upload an interview transcript to tidy the notes. No bad intent, just a Friday deadline. The result is client data sitting in tools the firm never vetted, cannot audit, and does not control. This is also where that 19% figure from the BCG study bites: a consultant leaning on an unvetted tool for a task it cannot handle produces confident, wrong work with the firm's name on it.
Why the default cloud answer falls short
A standard public AI service does not fix this, and for consulting the problems are sharper than in most industries. NDAs and engagement letters routinely restrict where data may go and who may process it, and sending it to a third party model rarely fits those terms. European clients now ask about data residency in the RFP itself, because a US hyperscaler is subject to the US CLOUD Act wherever the server sits, which sits awkwardly with GDPR obligations. Clients expect assurance that their information will never train a commercial model and never resurface in an answer for a competitor. When a recommendation is challenged, the firm has to show how the answer was produced and which sources it relied on, which most consumer tools cannot do. And a firm that builds its knowledge base and workflows on a platform it does not control has handed its competitive edge to a vendor.
Hence the paradox: the AI tools that are easiest to adopt are the hardest to defend in front of a client.
What sovereign AI for consulting looks like
The requirements are not exotic. They are the questions that already appear in client security questionnaires and procurement reviews, and three of them do most of the work.
The first is zero training on client data. A contractual commitment, enforced by the architecture, that client and firm data is never used to train models. Not by default and not in the fine print.
The second is real data sovereignty. Deployment on infrastructure inside the EU and under EU jurisdiction, with hard separation between one client's data and the next.
The third is traceability by design. Logging and access control built into the platform, so that answering a security questionnaire means showing the system rather than writing a promise.
Notice there is no trade-off on that list. Protecting client trust is not what using AI costs you. Done properly, it is the reason clients will let you use it at all.
Client trust is the new consulting currency
AI will change how consulting is priced, staffed, and delivered, and that shift is already underway. Whether a firm benefits from it comes down to one thing: whether clients trust what happens to their data.
Consulting leaders no longer get to decide whether their firm adopts AI. Their people already have. What they still get to decide is whether that happens in the open, on infrastructure the firm controls, or in the shadows.
GLBNXT builds sovereign, EU-hosted AI infrastructure for organisations that cannot afford to get data wrong. If your firm wants to scale its expertise without putting client trust at risk, talk to us.
Sources
Dell'Acqua et al., "Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of Artificial Intelligence on Knowledge Worker Productivity and Quality", Harvard Business School Working Paper 24-013 (with Boston Consulting Group): https://www.hbs.edu/faculty/Pages/item.aspx?num=64700
Business Insider, "AI is reshaping how McKinsey makes money" (McKinsey UK managing partner Michael Birshan, November 2025 briefing): https://finance.yahoo.com/news/ai-reshaping-mckinsey-makes-money-195132745.html
Consultancy.uk, "Is AI about to kill the billable hour consulting model?" (January 2026): https://www.consultancy.uk/news/42718/is-ai-about-to-kill-the-billable-hour-consulting-model
European Union, General Data Protection Regulation (EUR-Lex): https://eur-lex.europa.eu/eli/reg/2016/679/oj/eng
United States, Clarifying Lawful Overseas Use of Data (CLOUD) Act, H.R. 4943, 115th Congress: https://www.congress.gov/bill/115th-congress/house-bill/4943
