Garfield AI has moved from regulatory test case to live litigation example after helping a freelancer recover £7,000 in unpaid fees at Wandsworth County Court, a result that puts the Solicitors Regulation Authority’s approval of AI-led legal services under fresh scrutiny.
The claim was brought by Tamires Camal Taquidir, who used Garfield to pursue unpaid HR-related fees from a hospitality business after earlier attempts to resolve the dispute failed. Garfield prepared pre-action correspondence, issued the claim and produced the trial documents, including witness material and the court bundle.
Advocacy was handled by Dominic Li of One Essex Court, leaving the technology provider in the role that would traditionally have been performed by solicitors preparing the case for hearing.
Practitioner consequence now matters more than the novelty. This was not a robot judge, nor an AI advocate speaking in court, but it did place document preparation, case presentation and procedural guidance into a regulated technology workflow. For firms handling small debt claims, unpaid invoices and routine commercial disputes, the commercial question is whether clients will continue to pay conventional legal fees for work that can be packaged, automated and supervised at lower cost.
The SRA authorised Garfield.Law Ltd in May 2025 as the first purely AI-based firm permitted to provide regulated legal services in England and Wales. Its approval came with an important regulatory premise: solicitors remain accountable for the legal services delivered through the system. That distinction should not be lost as the market reacts to the court win.
The professional issue is not whether AI can draft a witness statement, but whether the person or firm responsible can prove the output has been checked, the client has understood the process and the court has not been misled.
The timing is sensitive for the wider profession. Kirkland & Ellis has said it will invest heavily in its own AI platform, while Pinsent Masons and Sullivan & Cromwell have both been drawn into separate concerns over AI-generated or AI-assisted errors in litigation filings. Garfield’s result therefore lands in a market split between enthusiasm for efficiency and anxiety about hallucinations, supervision and professional liability.
Philip Young and Daniel Long built Garfield around small debt recovery, an area where the cost of using lawyers can easily exceed the value of the claim. That access-to-justice argument is likely to carry weight, particularly if more claimants can recover modest sums without writing them off as uneconomic. The harder question for law firms is whether the same model can expand into other civil work without weakening the professional judgment that solicitors and barristers are expected to bring.
The SRA, Garfield AI and legal tech providers now have to show that regulated AI can reduce litigation costs while preserving solicitor accountability, court accuracy and client protection.