On 14 May 2026, at Wandsworth County Court in south London, something that looked routine on the surface turned out to be anything but. A freelancer won a debt claim worth £7,000. A counterclaim was dismissed. A judge delivered a reserved judgment. Barristers packed up and left. Standard fare for a small claims court.
Except that almost all the legal work that made that victory possible was done by a machine.
Garfield AI, the world’s first AI-powered law firm authorised and regulated by the Solicitors Regulation Authority (SRA), had handled every step of the litigation process leading up to the trial. It completed the drafting pre-action correspondence, preparing and issuing court proceedings, managing disclosure, producing witness statements, and assembling the complete trial bundle.
The claimant, freelance HR consultant Tamires Camal Taquidir, paid approximately £400 for all of that. The opposing side, by contrast, instructed both a solicitor and a barrister. The court found against them anyway.
Philip Young, co-founder of Garfield AI and a former City litigator who spent eight years at Baker McKenzie before co-founding his own litigation boutique, has called it “the dawn of a new age of access to justice.” It is a bold claim. It is also one that is very hard to dismiss.
The Problem This Solves
To understand why this case matters, it helps to understand the trap that millions of small businesses and freelancers have long found themselves in. When someone owes you money and refuses to pay, you can take them to court. In theory. In practice, the cost of hiring a solicitor to pursue a debt of a few thousand pounds can easily match or exceed the amount you are trying to recover. So, most people write it off and move on.
Garfield AI was built to break that deadlock. In just over a year of operation, the platform has processed more than 600 claims and recovered or resolved over £500,000 for its users, with claim values ranging from around £30 to £10,000.
Taquidir’s own account captures the reality of what a platform like this means for an ordinary person facing a legal dispute.
“I was owed money for work I had done, but it felt like the process of recovering it could be too stressful, expensive and time-consuming. Garfield made it possible for me to pursue the claim and keep going. When the counterclaim was brought, it was intended to intimidate me, but I knew I had accessible, cost-effective and competent support. I am delighted by the result,” she said.
That word “competent” is the one that should make traditional law firms pay attention. Competent support, for £400, against a defendant who had both a solicitor and a barrister in their corner.
What the Machine Actually Did
The AI did not stand up in a courtroom and address a judge. Oral advocacy at the hearing remained entirely human. Garfield instructed barrister Dominic Li of One Essex Court shortly before the trial began, and he presented the case in court.
What the AI handled was everything that came before, including the documents, the filings, the evidence preparation. In legal terms, that is called the pre-trial work, and it is also, historically, where a substantial portion of a solicitor’s fees accumulate.
Mark Lewis, a lawyer at Stephenson Harwood, acknowledged that the verdict validates the platform and shows that “used properly, and integrated into legal systems and court processes, this AI works as it should.”
Daniel Long, Garfield’s co-founder and chief technology officer, is careful not to overstate what happened.
“It is not about gimmicks or replacing lawyers. It is about giving people and businesses the tools to enforce their rights when the traditional route would be too slow, too costly or too complex,” he said.
The Threat That Is Already Arriving
According to the 2026 “Future Ready Lawyer Survey,” which covered 810 legal professionals across the United States, China, and eight European countries, more than 90% of lawyers already use at least one AI tool in their daily work.
McKinsey estimates that 44% of legal tasks are technically automatable with technology that already exists today. Corporate legal departments are adopting AI faster than the outside law firms they hire, and 64% of in-house legal teams now expect to depend less on external counsel because of AI capabilities they are developing in-house.
In February 2026, Baker McKenzie cut between 600 and 1,000 business services roles across know-how, research, marketing, and secretarial functions, citing AI integration. It was the largest AI-attributed reduction in legal industry headcount to date. Those were not lawyers, but support staff whose functions AI can now partially absorb.
The traditional structure involved hiring large classes of junior lawyers and assigning them high volumes of routine work, including document review, contract analysis, first-draft filings. That work justified the headcount and, crucially, it was also how junior lawyers learned their trade. AI tools are now handling much of that work faster and at a fraction of the cost.
The International Bar Association has flagged AI as a “critical issue” for the profession, noting that an increasing number of tasks previously carried out by young lawyers, interns, and trainees can now be handled relatively easily by AI, raising serious questions about how a legal career develops going forward.
The worry is not just about jobs disappearing. It is about whether the profession can still produce experienced senior lawyers if the foundational work that once built their skills no longer needs to be done by humans.
Where the Machine Falls Short
In a rigorous benchmarking exercise conducted by ScaleAI, the best-performing AI model scored only 37% on the most difficult legal problems, meeting just over a third of the possible evaluation criteria. That is not a passing grade by any standard.
Courts and regulators have issued repeated warnings that lawyers remain personally responsible for verifying everything they submit, after filings from major firms were found to include AI-generated references that turned out to be completely fabricated.
Garfield’s model works in part because it operates within a narrow, well-defined domain. Small claims debt recovery in England and Wales follows a relatively standardised process. The documents required are predictable. The legal questions involved, while real, are not typically the kind of complex, novel problems that expose the current weaknesses of AI reasoning.
Applying the same approach to criminal defence, corporate litigation, or constitutional law is a fundamentally different challenge, and nobody is seriously claiming that territory is within reach anytime soon.
The Bigger Picture
What is striking is that the largest law firms in the world are not waiting to find out whether AI will disrupt them. They are treating disruption as a given and racing to control how it happens. Kirkland and Ellis has committed $500 million to developing its own proprietary AI tools, while Freshfields has struck a partnership deal with Anthropic. These are offensive investments by institutions that intend to own the technology rather than be displaced by it.
Despite the disruption narrative, legal employment in the United States reached a 10-year high of 1.24 million jobs in January 2026, according to Bureau of Labor Statistics data. The profession as a whole is not shrinking. But it is changing shape, and the changes are moving fastest at the base of the pyramid, where junior lawyers and support staff do the high-volume, process-heavy work that AI is best equipped to absorb.
If a freelancer can recover £7,000 in unpaid fees for around £400 using an AI-driven platform, the logical question is how many clients will continue to choose a traditional firm for the same category of work.
What happened at Wandsworth County Court was not the death of the legal profession. It was proof that a machine, working within a regulated framework, can outperform an expensive human team on a real legal problem with real stakes. The solicitors and barristers of Britain might reasonably conclude that a £7,000 small claim is beneath their concern. History suggests that is exactly the kind of reasoning that tends to age badly.