Lean Over Licences: Boxtree's Process-First Model Bets Against the AI Hype Cycle

The pitch is blunt enough to stand out. Max Pardo-Roques, previously Commercial Director at specialist Lean firm Develop Consulting, launched Boxtree Consulting earlier this month with a single provocation: most British businesses think buying AI software is the same thing as implementing AI strategy, and the gap between those two things is where consulting fees go to die.

Boxtree, based in Birmingham's Jewellery Quarter, applies Toyota-derived Lean methodology to map administrative processes before any AI solution is specced or procured. A diagnostic puts a pound value on waste inside each workflow; a business case is built per opportunity; implementation only proceeds where first-year returns provably exceed the cost of the work. The firm is targeting manufacturers, logistics businesses, and professional services clients, and is guaranteeing every engagement is cost-neutral within year one, with a stated 5:1 ROI target.

"Look closer and they've bought ten or fifteen Copilot licences, a few people are writing better emails and the marketing copy has sharpened up," Pardo-Roques said in a launch statement. "That's individual productivity, not transformation. Nobody has touched the processes where the money actually sits."

The process-first sequencing is a deliberate inversion of the standard SaaS-led approach. It's also, frankly, a harder sell at board level, where AI spend is often driven by competitive anxiety rather than operational diagnosis. But the numbers behind Boxtree's launch rationale are credible. Official UK government research published in February 2026 found that only 16% of UK businesses are currently using at least one AI technology, even as senior leaders publicly signal commitment to AI transformation. The Management Consultancies Association has separately found that 77% of UK consulting firms have integrated AI into their own internal systems, yet the same firms are watching clients stall at pilot stage.

The Talent Crunch Underneath

Boxtree's formation coincides with a structural problem that makes process-led consultancy look even more rational. Research published by freelancer platform Malt this week found that demand for AI agent expertise increased by 5,800% over the past year, while the number of AI engineer freelancers on its platform grew by only 229%. That is a supply-demand mismatch of unusual severity. Malt also noted AI is now the second most in-demand skill across its million-strong freelancer network, regardless of sector or company size.

For a firm like Boxtree, the talent gap actually strengthens the proposition. If specialist AI engineers are scarce and expensive, the commercial logic for a Lean diagnostic that identifies the fewest, highest-return automation targets becomes sharper. You don't need an army of MLOps specialists to automate an invoice-checking workflow; you need a clear process map and someone who understands where the waste is. Boutique AI consultancies are already capturing increasing mid-market share precisely because they build directly on existing AI platforms rather than adding project management overhead that inflates cost and extends timelines.

Formation Data Points to a Crowded Market

The timing is not without risk. AI Business Dispatch analysis of Companies House data shows 2,152 new SIC 70.22 companies formed in 2026-Q3, a figure down 83.2% on the prior period as the initial formation wave corrects. That contraction cuts both ways: the cohort is thinning, but so is the noise. The firms that formed in the 2024-2025 rush and cannot demonstrate commercial traction are already showing attrition signs.

On the intellectual property side, the same analysis of IPO trademark data records 1,946 Class 41 filings in Q3 2026, down 77.1% on the prior period. More telling: 98.6% of active SIC 70.22 companies hold no Class 41 trademark at all. For a consultancy sector that trades on proprietary methodology and repeatable frameworks, that is a striking degree of brand underprotection. Boxtree's Lean-derived diagnostic process could, in theory, represent exactly the kind of protectable IP that most sector entrants are leaving unguarded. Source: AIBD analysis of Companies House (MC) / IPO (TMD) data, as of 2026-07.

The Compliance Overhang

For recruitment-adjacent consulting clients, there's a parallel pressure building. The Data (Use and Access) Act 2025 came into force on 5 February 2026, rewriting the rules on automated decision-making. The ICO's March 2026 report, drawn from evidence from more than 30 UK employers, found that most organisations using AI screening tools were effectively making solely automated decisions while describing them internally as decision support. Meaningful human involvement, the ICO made clear, means the reviewer must have the authority, discretion, and competence to change an outcome before it takes effect. Rubber-stamping an AI shortlist does not qualify.

That compliance burden is arriving just as AI-powered recruitment tools are proliferating fastest. For consultancies advising HR functions or running their own talent pipelines through automated screening, the regulatory gap between current practice and ICO expectations is not small.

Boxtree won't fix that problem directly. But its underlying logic, start with the process and the regulation rather than the technology, applies equally to recruitment automation as to invoice processing. The firms that survive the current consulting cohort correction will likely be the ones that asked the boring operational questions first.