An AI law firm in England has claimed a world-first by winning a contested trial.
In May at London’s Wandsworth County Court, an HR consultant used Garfield AI’s automated system to recover about £7000 she was owed by a hospitality business.
The firm’s AI-powered litigation assistant generated the client’s pre-action correspondence, issued the claim, produced the documents, prepared the witness statements, then briefed a human barrister.1
After a three-hour contested hearing, the court found for her in full and dismissed the defendant’s counterclaim.2
In May last year the UK’s legal watchdog, the Solicitors Regulation Authority, authorised Garfield.Law Ltd as the first purely AI-based firm permitted to provide regulated legal services in England and Wales. Several other examples have followed, all aimed at high-repetition-low-fee work which a full-service firm would struggle to undertake cost-effectively.
The authorisation came with conditions: the system takes no step without client approval, it is configured not to cite case law (where large language models most often invent authority), and named solicitors remain accountable for its output, backed by mandatory professional indemnity insurance.3
Garfield’s technology blends Large Language Models (to manage the human-interaction and bespoke drafting) with more structured “expert systems” relying on fixed templates and precedents to produce most documents. While far from risk or error free, this kind of hybrid structure applied to a strictly limited range of legal problems makes a highly automated production system more viable.
Would a system like this fit within Queensland regulatory landscape?
To an extent, yes. Provided the firm is a properly constituted Law Practice with (if applicable) a legal practitioner director4 who ensures appropriate supervision of the firm and the firm’s work, there is no theoretical limit on the degree of automation a firm can apply to production of legal work.
In practice, Queensland is not ready for robo-lawyers. The level of supervision required to properly acquit a legal practitioner’s duty of competence, duty of professional independence and duties to the Court5 would make it difficult for a firm to supply claim and response materials without a human in the loop, other than as a self-service limited retainer.
The automated system – and certainly preparation by an advocate at trial, would require a human with sufficient knowledge of the facts and law to review the evidence and draft material before supplying it to the client.6 While Practice Direction 5 of 2025 does not apply to pre-trial correspondence or pleadings, fundamentally that practice direction is declaratory of a lawyer’s existing professional obligations; if you sign or present work to the Court you accept responsibility for it and warrant that you have taken reasonable steps to ensure that a claim can be advanced.7
Why this is still a significant milestone
England has gone further than we have in licensing AI in law, and the general acceptance expressed by judges and the government of the UK demonstrates greater enthusiasm for the potential of automation to enhance access to justice.
While Australian Courts and regulators certainly preface most of their statements by acknowledging AI’s potential to make legal systems more equitable8 it is reasonable to say that caution and focus on errors remains the predominant theme, especially in NSW.
For Garfield AI’s client, the reality was probably that without a firm that could prepare her case, respond to a counterclaim and instruct counsel for (reportedly) £400 she would have had to walk away from her payment. A carefully structured and implemented process can be consistent with our professional ethics. No doubt there will be many mis-steps and mistakes along the way, and a relatively straightforward debt claim is not the same as more complex novel litigation, but for us the lesson is clear: adapt or see the work pool shrink inexorably year by year as the sophistication of automated systems improves.
Footnotes
1 The degree to which human solicitors or paralegals supervised and participated in these processes behind the scenes is not known. See the firm’s own account on its website: Garfield AI, ‘AI Lawyer in Historic Wins First Court Trial (Web Page, 22 June 2026) <https://www.garfield.law/press/garfield-ai-wins-first-court-trial-with-regulated-ai-lawyer>.
2 As at 23 June 2026, no judgment in the matter appears on the National Archives’ Find Case Law service (https://caselaw.nationalarchives.gov.uk) or on BAILII; County Court small-claims decisions are seldom transcribed or reported, and the claim number has not been made public. Facts are sourced from UK media reporting: Charlie Moloney, ‘HR consultant wins English court case using AI lawyer in apparent legal first’, The Guardian (Sydney, 23 June 2026); Anthony Cuthbertson, ‘How an AI lawyer won a UK court case in ‘legal first’, Independent (London, 23 June 2026); Scott Jones, ‘First court case win for AI-powered law firm’, Legal Futures (23 June 2026); and the firm’s website (see n. 1).
3 Solicitors Regulation Authority, ‘SRA Approves First AI-Driven Law Firm’ (News Release, 6 May 2025). Details on the agreed risk management processes underlying the approval were not made available publicly.
4 Legal Profession Act 2007 (Qld) s 117.
5 Queensland Law Society, Australian Solicitors Conduct Rules 2023 (ASCR), R.r 4.1.3, duty to exercise independent judgment, ASCR r 17.1 and duty not to mislead the Court ;, ASCR r R.19.
6 See Queensland Law Society, Artificial Intelligence in Legal Practice (Guidance Statement No 37, 31 May 2024
7 Supreme Court of Queensland, Practice Direction No 5 of 2025: Accuracy of References in Submissions, 24 September 2025, 1–3.
8 For example, Supreme Court of Victoria, Practice Note SC Gen 25: The Use of Artificial Intelligence by Court Users; (14 May 2026).
Share this article