How will AI cut Headcount: You're Asking the Wrong Question

The dominant conversation around AI at work is about replacement. Will it take jobs? How many? How soon? It's the wrong question, and organisations fixated on it are not only missing the point, they are missing the opportunity.

Using AI primarily as a headcount-reduction tool is a failure of imagination. It treats a transformative technology as a cost-cutting exercise, extracts short-term savings, and leaves the real value on the table. The organisations that will win are those asking a different question entirely: what can we now do that we couldn't do before?

No example makes this clearer than Garfield Law.

The Garfield Law Story

In May 2025, the Solicitors Regulation Authority (SRA) made history by authorising Garfield.Law Ltd as the first purely AI-driven law firm in England and Wales. Based in Tunbridge Wells, it helps individuals and small businesses recover unpaid debts through the small claims court — guiding users through every step from the initial chaser letter to trial preparation.

The founding insight was simple and powerful. Co-founder Philip Young, a commercial litigation lawyer, watched his brother-in-law — a plumber owed money by a client — give up on pursuing a legitimate debt claim. Not because he had no case, but because the legal system was too slow, too expensive, and too opaque for someone in his position.

That work, small debt claims, was never taken from anyone by AI. It was work that wasn't happening at all. Traditional law firms couldn't serve it profitably. Individuals couldn't afford it. The market existed; the economics didn't. Garfield made them work.

The platform now guides users through the entire claims process: validating the claim, generating pre-action letters, drafting court forms, managing defences and counterclaims, and preparing for trial — all for a fraction of traditional legal costs. A Channel 4 investigation found Garfield producing court-ready documents for around £100 plus VAT, against roughly £1,080 for a conventional solicitor. More significantly, for many users, the alternative wasn't a solicitor. It was nothing.

What This Means for Every Organisation

Garfield didn't replace lawyers. It served a market lawyers had never reached. That is the question every leadership team should be asking about AI: where is our equivalent of the unpursued debt claim?

Every organisation has work that doesn't get done — not because it isn't valuable, but because the economics never stacked up. Customer enquiries that go unanswered after hours. Analysis that would take a team a week to run. Services priced out of reach for smaller clients. Internal processes so labour-intensive that no one has ever properly tackled them.

AI changes the unit economics of that work. The opportunity is not to do the same things with fewer people. It is to do things that were previously impossible, impractical, or unaffordable — and in doing so, reach new markets, serve more customers, and create genuine value.

Human Oversight Is Non-Negotiable

None of this means AI operates without accountability. The SRA's approval of Garfield came with clear conditions: every action requires explicit client approval, named regulated solicitors remain ultimately responsible for professional standards, and the system is deliberately prevented from citing case law — one of the highest-risk areas for AI errors. Human oversight is not a workaround; it is built into the design.

The best AI deployments across every sector follow the same principle: AI handles the structured, high-volume, process-driven work. Humans provide judgement, accountability, and the relationships that technology cannot replicate.

The Right Question

The firms and organisations that thrive will not be those that used AI to trim headcount in 2025. They will be those that used it to unlock markets, serve customers they couldn't previously reach, and build things that simply weren't possible before.

Garfield Law is a small firm handling unglamorous cases. But it is also proof of what happens when you stop asking "how do we do the same work cheaper?" and start asking "what work can we finally do?"

That is the question worth answering.

Sources: Solicitors Regulation Authority; Law Society of Scotland; Channel 4 Dispatches; Computer Weekly.