The AI Licence Illusion: Why UK Consulting's Spend-and-Hope Model Is Running Out of Road

The pitch from Boxtree Consulting, launched this month from Birmingham's Jewellery Quarter, is uncomfortable enough to be credible. According to founder Max Pardo-Roques, most UK businesses have "AI implementation backwards": they purchase licences, circulate login credentials, and consider the transformation complete. Boxtree's model applies Lean methodology, the same discipline derived from the Toyota Production System that reshaped manufacturing in the 1990s, to find the actual manual processes that AI can automate, building solutions inside the systems clients already run rather than layering new subscriptions on top.

Pardo-Roques, who spent more than 15 years in operational excellence and transformation consultancy and served most recently as commercial director at specialist Lean firm Develop Consulting, draws a sharp historical parallel. "Twenty-five years ago, businesses responded to Lean by sending junior staff on training courses, and little changed until specialist implementation firms got involved," he said in the firm's launch press release. "AI is at exactly that point. The licences are the training course. Implementation is where the value is."

The numbers behind that claim are hard to argue with. ONS data analysed by the Bennett School of Public Policy shows that fewer than three in ten UK businesses with under 50 staff were using AI in 2025, against 44% of firms employing more than 250. That is not an adoption gap so much as a delivery failure, and it is precisely the territory that a Lean diagnostic, which maps a client's processes and assigns a pound value to the waste within them, is designed to expose.

Boxtree promises clients a 5:1 ROI and what its founder calls a "cost-neutral" Year 1. Deliberately unglamorous, as Pardo-Roques puts it. Unglamorous is exactly what the formation data suggests the market needs right now.

Formation Data Tells an Inconvenient Story

AI Business Dispatch analysis of Companies House and IPO data complicates the consulting boom narrative. New SIC 70.22 (management consultancy) incorporations in 2026-Q3 stand at just 2,719, a 78.7% fall versus the prior period. Class 35 UK trademark filings, the broadest proxy for brand intent across business services, have dropped 71.7% in the same quarter to 3,060 registrations. Most striking: 98.2% of active SIC 70.22 companies hold no Class 35 trademark whatsoever. That statistic reveals something about how most UK management consultancies are built, on reputation, relationships, and no formal brand infrastructure at all. (Source: AIBD analysis of Companies House / IPO data, as of July 2026.)

The formation slump does not mean the consulting market is contracting. It means the easy-entry phase, two people, a laptop, and a LinkedIn headline proclaiming AI expertise, may be closing. The Management Consultancies Association found that 77% of UK consulting firms have integrated AI into their systems or enabled employees to use AI models, with 76% deploying AI for research tasks and 68% increasing automation. Official UK government research published in February 2026 found that only 16% of UK businesses are currently using at least one AI technology. The gap between consulting supply and client readiness is significant.

The Recruitment Compliance Cliff

For the SIC 78.20 recruitment sector, an adjacent pressure is crystallising. The ICO's final guidance on automated decision-making is due this summer, following a consultation that closed on 29 May 2026. The stakes are not theoretical.

The ICO's March 2026 report, which drew on voluntary engagement with more than 30 UK employers between March 2025 and January 2026, found that most are non-compliant with automated decision-making rules. The central finding: employers reported that their AI tools were used only for decision support, with a human making the final call. The evidence showed something different, tools making substantive decisions and human review that amounted to rubber-stamping. The ICO has since written formally to sixteen named organisations, which have committed to act.

The Data (Use and Access) Act 2025, which came into force on 5 February 2026, replaced the old near-total prohibition on significant automated decisions with a permissive, safeguard-led regime. UK agencies can now use AI screening tools under legitimate interests, but only with proper safeguards: transparency, meaningful human review, and the right for candidates to contest decisions. The regulator is explicit that approving an AI-generated shortlist without genuine capacity to override it does not qualify as meaningful involvement.

For recruiting consultancies and in-house HR functions that built workflows around AI screening during the past two years, the compliance ask is now concrete and imminent. Every tool that filters, ranks, or rejects candidates needs a Data Protection Impact Assessment, documented bias monitoring, and a reviewer with actual authority to change an outcome. Vendor due diligence in writing, as practitioners note, because the vendor's compliance becomes your compliance.

Two Pressures, One Structural Conclusion

Put Boxtree's argument alongside the ICO's enforcement posture and you get the same diagnosis from two directions. Firms that deployed AI for speed, in consulting delivery or in candidate screening, without rethinking the underlying process are now exposed on both counts: commercially, because the ROI is not materialising, and legally, because the regulator has spent two years gathering the evidence to act.

The Thomson Reuters Institute's 2026 AI in Professional Services Report found that organisation-wide usage of AI in professional services has almost doubled in the past year to 40%, compared to 22% in 2025. Usage is not the same as value, and value is not the same as compliance. The firms that survive the next 18 months in this cohort will be those that can show their working, on implementation, on governance, and on what a human reviewer actually does when the algorithm recommends a shortlist or a strategic option.

Buying the licence was the easy part.