An AI litigation assistant prepared the entire case file. A human barrister argued it in court. The judge ruled for the claimant. That sequence is now a matter of public record, and it happened under a regulatory framework that most compliance professionals haven’t examined closely yet.
Tamires Camal Taquidir instructed Garfield AI (operating as Garfield.Law Ltd) to pursue a £7,000 debt recovery claim through Wandsworth County Court’s small claims track. The AI system drafted the witness statements and trial bundle. A human barrister handled the hearing. The judge awarded the full £7,000 and dismissed the defendant’s counterclaim. Garfield AI charges as little as £2 for a pre-action letter and £50 to file a claim, making the total cost reportedly around £400, consistent with the firm’s published pricing.
The SRA authorization is the part that matters here. The Solicitors Regulation Authority confirmed that Garfield.Law Ltd is the first purely AI-based firm it has authorized to provide regulated legal services in England and Wales, having received that authorization in 2025. That authorization isn’t symbolic. It required Garfield AI to meet the same regulatory obligations as any other SRA-regulated firm: professional indemnity insurance, client money rules, and the SRA Standards and Regulations. The AI system isn’t operating outside the regulatory perimeter, it’s inside it.
The AI-Human Division of Labor
The access-to-justice dimension is real. Conventional small claims legal assistance typically costs multiples of the £400 reportedly charged here. For debt recovery disputes under £10,000, exactly the range Garfield AI currently serves, the economics of conventional legal help often make pursuing a claim irrational. Founder and CEO Philip Young described the outcome as “a landmark moment… for access to justice.” That framing is plausible, though what it means for regulatory architecture deserves more scrutiny than the press coverage has given it.
The catch is the AI-human division of labor. Garfield AI handled pre-trial preparation; a human barrister reportedly argued the case in court. That structure isn’t incidental, it reflects current SRA guidance on AI’s role in regulated legal services. An AI system that prepares an airtight case file doesn’t face the same regulatory exposure as one that appears before a judge. This trial doesn’t test the limits of what an AI litigation assistant can do in a courtroom. It tests the limits of what it can do before the courtroom, and that boundary held.
Who This Affects
Don’t expect this to generalize quickly to complex litigation. Small claims track disputes are procedurally bounded, fact-intensive rather than law-intensive, and resolved without jury deliberation. The regulatory architecture that permitted this outcome was designed for exactly this scope. What happens when a similarly authorized system attempts a more contested or legally complex matter remains an open question, one the SRA will almost certainly face before long.
The real question is whether other professional services regulators are watching. The FCA, the GMC, and analogous bodies across financial advice and medical professional services each face equivalent authorization questions as AI tools move into regulated professional workflows. The SRA’s 2025 authorization of Garfield AI is the first answer on record. The compliance community should be studying what that authorization actually required, not just what the court outcome achieved.