Garfield.Law Ltd

Small and medium-sized enterprises pursuing unpaid debts under £10,000 face a structural gap in England and Wales: traditional legal representation is economically unviable at this scale. Solicitors cannot profitably handle high-volume, low-value small claims disputes, leaving businesses — particularly those with recurring trade debtors — effectively unable to enforce payment without absorbing the cost themselves. Filing in the small claims track without legal help carries procedural risk, while hiring a firm typically costs more than the debt recovered. The result is a large segment of commercial disputes that go unpursued, creating a systemic access-to-justice failure for the SME market.

Garfield.Law built a domain-specific litigation platform powered by Large Language Models and Generative AI, purpose-designed to handle every procedural step of the English small claims process end-to-end. The system assesses claim validity, verifies limitation periods, generates pre-action correspondence, drafts and submits claim forms, manages defences and counterclaims, compiles trial bundles, and supports settlement negotiations — all within a structured, rule-constrained workflow that reduces hallucination risk by narrowing the AI's scope to a single procedural domain. Crucially, client approval gates each step before action is taken, and an experienced litigator supervises the platform during its initial operating phase. This supervised-autonomy architecture allowed Garfield.Law to seek and obtain formal regulatory authorisation rather than operating outside existing frameworks.

On 6 May 2025, the Solicitors Regulation Authority authorised Garfield.Law as the first purely AI-driven firm approved to provide regulated legal services in England and Wales — a regulatory milestone with no precedent in the jurisdiction. The platform's cost structure reflects the efficiency of AI-at-scale delivery:

• £2 per gentle pre-action chaser letter
• £7.50 per formal Letter Before Action

These prices place debt recovery within reach for SMEs that previously had no economically rational legal option for claims under £10,000. The SRA authorisation also validates the firm's compliance architecture, confirming that AI-driven legal services can meet professional conduct standards when properly supervised.

• Regulatory authorisation is achievable for AI-driven legal services when the system is built within the framework from inception — Garfield's SRA approval demonstrates that supervised autonomy satisfies professional conduct requirements.
• Domain specificity reduces risk: constraining LLM outputs to a single procedural area (small claims debt recovery) enables rule-based precision and lowers the likelihood of legally consequential errors.
• Extreme price compression (£2–£7.50 per letter) is a signal, not an endpoint — AI legal services compete on scale economics, not traditional hourly margins.
• Human oversight at the deployment stage is not a concession but a prerequisite; litigator supervision during early operation is what made regulatory approval possible.