AI Licences Are Not a Strategy: Boxtree's Lean Bet on the UK Consulting Gap
A Birmingham boutique is staking its entire model on the claim that most UK businesses have AI implementation backwards. The evidence, from Companies House to Canary Wharf, suggests it may be right.
The Licence Delusion
Buying a software seat is not the same as changing a process. That distinction has always been obvious to anyone who has watched an ERP roll-out go sideways, but it appears to need restating for the AI era.
Boxtree Consulting, launched this week from Birmingham's Jewellery Quarter, is making that argument its entire commercial proposition. Founded by Max Pardo-Roques, formerly commercial director at specialist Lean improvement firm Develop Consulting, the firm applies Toyota Production System methodology to identify which manual processes AI can realistically automate and then builds the solutions inside the systems clients already operate. The target client is not a FTSE 100 innovation lab. It is the finance director of a mid-market manufacturer who is paying for Copilot licences and cannot point to a single process that runs differently because of them.
Advertisement
Pardo-Roques frames the problem in terms that will be familiar to anyone who sat through the first decade of Lean: "Twenty-five years ago, businesses responded to Lean by sending junior staff on training courses, and little changed until specialist implementation firms got involved. AI is at exactly that point. The licences are the training course. Implementation is where the value is."
Boxtree's model starts with a Lean diagnostic that maps a client's processes and puts a pound figure on the waste inside them. A business case is built for each opportunity before any technology decision is made. The firm claims clients can expect a 5:1 return on investment, with Year 1 described as cost-neutral by design.
The Structural Problem Behind the Launch
Boxtree's timing is not accidental. The gap between AI spending and AI returns has become one of the more uncomfortable facts in the UK economy. Fewer than three in ten UK businesses with under 50 staff were using AI in 2025, against 44% of firms employing more than 250, according to ONS data analysed by the Bennett School of Public Policy. Separate government research published in February 2026 put active AI usage across UK businesses at just 16%. The consulting sector itself reflects this ambivalence: the Management Consultancies Association found that 77% of UK consulting firms have integrated AI into their systems or enabled employees to use AI models, yet measurable P&L impact remains elusive for most.
The supply side is tighter still. A report published this month by freelancer platform Malt found that demand for AI agent expertise increased by 5,800% over the past year, while the growth of those skills in the UK workforce is running at roughly half that rate. AI is now the second most in-demand skill across Malt's platform regardless of company size, industry or project type. The number of AI engineers on Malt's network of one million freelancers increased 229% in 12 months, but that is still not enough: major UK organisations are turning to independent AI specialists and fractional executives to bridge skills shortages and support transformation programmes.
For a boutique like Boxtree, that talent scarcity is both a threat and a competitive moat. The firm is not selling AI engineering. It is selling operational expertise wrapped around AI tools, a meaningfully different skill set and one that is harder to commoditise.
What the Formation Data Says
The launch arrives against a notable shift in the consulting formation market. AI Business Dispatch analysis of Companies House data shows 2,312 new SIC 70.22 (management consultancy) companies registered in 2026 Q3, down 81.9% on the prior period. A surge of new entrants has clearly peaked. The boutique AI consultancy boom that characterised 2024 and early 2025 appears to be consolidating, and the firms that survive will need a defensible positioning. Lean-AI methodology, with its emphasis on process mapping before tool selection, is at least a coherent answer to the ROI question that most new entrants have been unable to answer.
The brand picture is equally revealing. IPO trademark data shows only 420 Class 45 filings in 2026 Q3, down 73.5% on the prior period. More striking: 99.7% of active SIC 70.22 companies hold no Class 45 trademark at all. For a sector selling trust, judgment and proprietary methodology, the near-total absence of formal brand protection is a structural vulnerability. A firm that cannot protect its name cannot defend its positioning when a larger rival decides to imitate it. Source: AIBD analysis of Companies House (MC) and IPO (TMD) data, as of July 2026.
The ICO Wrinkle for Recruitment-Adjacent Consultancies
One area where Boxtree's clients will need careful guidance is recruitment process automation. The ICO's March 2026 report, drawing on evidence from more than 30 UK employers, found that most organisations using AI to screen and score candidates are non-compliant with automated decision-making rules. The core finding was blunt: employers told the regulator their AI tools were used only for decision support, with humans making the final call. In practice, tools were making substantive decisions and human review amounted to rubber-stamping. Final ICO guidance is expected this summer and enforcement follows. Any operational AI deployment that touches hiring, performance scoring or workforce allocation now carries compliance risk that a Lean diagnostic alone will not surface.
The Unglamorous Bet
Pardo-Roques has described Boxtree's model as "deliberately unglamorous." That is probably the most honest positioning in UK AI consulting right now. The glamorous end of the market, staffed by ex-McKinsey partners selling agentic AI roadmaps at £300,000 a project, has a congestion problem. The unglamorous end, working with the invoice-checking and scheduling processes of a Birmingham logistics firm, does not.
Whether the 5:1 ROI claim survives first contact with a live client book is a question the next 18 months will answer. The structural argument, at least, is sound. Lean implementation succeeded not because the methodology was novel but because someone finally showed up to do the hard work of changing the actual process. AI is waiting for the same moment.