Agentic AI in the justice system: getting the foundations right

On 5 October 2026, the Law Society published a foresight report considering how autonomous AI (referred to as "agentic") AI could reshape the justice system in England and Wales. The report suggests that the profession is approaching an important point. There is still an opportunity to influence how agentic AI becomes embedded within the justice system - but that opportunity may be temporary. For businesses involved in litigation, the implications go beyond how quickly the courts adopt new technology. Decisions being made now about infrastructure, data, governance and human oversight could shape how effectively AI supports the justice system in future.

The pace of development  

Three years ago, generative AI was a curiosity for most UK lawyers. In July 2023, just 11% were using it for work. Today, LexisNexis research finds that 94% of UK legal professionals use AI for legal work, with nearly three-quarters using it at least weekly and a third every day.

The courts, businesses and lawyers have tried to keep pace with it. The judiciary has updated its AI guidance three times since December 2023, and on 9 October 2026 confirmed a cautious but positive approach: approved tools only, and no AI deciding any case.

The direction is consistent: AI is welcome, but responsibility and judgement remain human.

Why the Law Society's justice report matters

The future of agentic AI in the justice system is the most thorough look yet at what agentic AI could mean for courts, tribunals and dispute resolution in England and Wales.

Its headline finding is that agentic AI will not fix a failing system; it will compound it. And it describes the Law Society's own role as shifting from guidance to boundary-setting: deciding not just how AI is used, but what it may do and where human judgement, oversight and accountability must be retained. For the justice system, that makes the foundations for AI as important as the technology itself.

The six key takeaways below explain why.

Six key takeaways

• Digital reform has a long way to go. HMCTS has spent over £1 billion on digital transformation, yet parts of the process remain on paper, and the system still handles more than eight million paper forms a year.
• Fragmented platforms limit progress. Lawyers work across Money Claims Online, a legacy system, Possession Claims Online, the Damages Claim Portal and Online Civil Money Claims (OCMC). OCMC shows promise but is too narrow, with too many ways to drop "offline": more than one unrepresented defendant, or an opponent's solicitor not registered on MyHMCTS, is enough. Many Consumer Credit Act claims fall outside it altogether.
• Clean data and paper-free courts come first. There is broad agreement that agents fed poor data multiply their mistakes, and the report insists that basic digital infrastructure must be in place before agentic tools arrive. Most law firms went paperless years ago. However, in the family court, litigants in person still post documents to a bulk scanning centre. Until the courts catch up, inefficiency is built in.
• Get intake, triage and tracking right. The potential applications of agentic AI with the widest support are those that address administrative inefficiencies: checking filings are complete, routing cases and listing.   Users need to know that a filing has been received and what processing times look like to avoid clogging up the court system with enquiries about updates. It is clear that the current ways of contacting a county court simply do not work.
• Find the money for the foundations. No meaningful conversation about AI in the courts can happen until the problems identified are funded: poor data, paper systems, fragmented agencies and no fallback when systems fail. Agentic AI could dramatically reduce the administrative burden, but only if it is deployed carefully, incrementally and with robust human oversight.
• Design for vulnerability from the outset. If AI becomes the front door to justice, those who cannot use it risk being locked out entirely. FCA regulated firms are already expected to identify vulnerable customers, and the courts should do the same at the first point of contact.

What this means for you

The Law Society identifies a narrow but significant opportunity to influence how agentic AI may be embedded in the court system, and warns that this opportunity may be short. The transformation is likely to unfold gradually, meaning that decisions made today about the procurement, infrastructure, governance and deployment of AI will have lasting consequences for how the technology operates within the justice system.

It is therefore critical that businesses that receive or are involved in higher volumes of litigation engage actively in the discussions now. The time is now to influence how the court systems could be improved. There is a genuine opportunity to help shape a court system that uses AI to improve access and efficiency, while retaining the human judgement and accountability on which confidence in the justice system depends.

Talk to our disputes team about AI and technology in litigation.

Contributors: Jordon Bray, Alanna Tregear, Victoria Channing, Amy McConnell, Amy Earlam, Helen Carter

This publication is intended for general guidance and represents our understanding of the relevant law and practice as at October 2026. Specific advice should be sought for specific cases. For more information see our terms & conditions.