Legal Geek Conference 2026
October saw the eleventh edition of Legal Geek’s flagship conference, its first since it was acquired by Centellic last year. I was interested to see what had changed. This was the biggest ever edition with 5,500 attendees from over 65 countries. The registration queue was over an hour, followed by queues for the cloakroom and coffee, where someone said to me this was the most queues they had seen at any conference. Fortunately, the big crowds also meant a massive party vibe.
The content was more blended – i.e. there were more interactive sessions, and networking facilities. The overarching theme was still AI, but the mood has moved on from the feverish AI evangelism of recent years to a genuine interest (not confined to decision makers, AI developers and vibe-coders) in using AI to improve efficiency and quality, and avoid AI slip-ups. Law firms and corporate legal departments are seriously interested in getting the best from their AI investments. Which AI tools and platforms are the best fit, and what skills do you need to make a difference? One of the hot topics is agentic AI, which also raises legal and ethical questions, including what is best delegated to AI, and what still needs human oversight and judgment.
Business class upgrade
Legal Geek’s party vibe got a business class upgrade with vendors, firms and institutions splashing out on massive stands, glamorous hospitality areas and extravagant ‘swag’. Harvey had a massive corner stand and Legora, which on 21 October raised a $550m Series E funding round raising its valuation to $8.8bn, had its own exclusive hospitality lounge. Not to be outdone, Thomson Reuters’ CoCo lounge hosted a meet and greet with the day 2 keynote speaker Louis Theroux. The massive funding now going into legal AI has shifted Legal Geek from a hipster start-up vibe to something resembling the big US tech trade shows.
A slippery slope?
One Legal Geek tradition that didn’t change this year was the opening keynote speech from the outgoing Master of the Rolls, Sir Geoffrey Vos. As he was into his last two weeks in office, he was more forthcoming, shifting from the AI evangelism of his previous Legal Geek keynotes to raising pertinent questions about AI in judicial decision-making. His 2024 prediction that lawyers and judges who refused to adopt AI would be left behind, has become a complete reality he said. Now that lawyers routinely consult LLMs and use legal AI platforms, judges in England and Wales have a ring-fenced version of Microsoft CoPilot and judges in other jurisdictions use AI to help them decide cases, Sir Geoffrey expressed concern about the effect on judicial decision making as it becomes harder to articulate a boundary between routine tasks and tasks that are part of decision making. He suggested “we may be on a slippery slope” and underlined the importance of continuing the discussion, notably through the Standing International Forum of Commercial Courts (SIFoCC), which is attempting to distil good practice in the use of AI and addressing the risks of reducing judicial involvement in deciding cases. Sir Geoffrey has resolutely worn a suit and tie to Legal Geek, and his combination of traditional attire (and a traditional role in the judiciary) with a sharp focus on the future of law brought a unique perspective to legal’s AI evolution.
The Legal Geek experience
There followed a series of highly professional TED-style lightening talks. Originally these sharply differentiated Legal Geek’s hipster vibe from conventional industry events. Now, they are just one small element of the Legal Geek experience. This year there were multiple ways to connect and share insights. A variety of interactive/alternative sessions helped maintain the Legal Geek energy, which was high as ever, even after 11 years!
I decided to dip in and experience a selection and queued up for Addleshaw Goddard’s popular Agentic Bloomprint Workshop on how to grow your own AI agent. Helpful session leaders talked us through designing, creating and refining an AI agent to handle a specific legal process. I also appreciated Addleshaw Goddard’s WeWork style lounge, where you could plug in your laptop and catch up with work.
A roundtable discussion led by Christina Blacklaws covered trust in the age of AI and the role of meaningful human oversight and human judgment. We also talked about AI governance and broadening access to the profession. Again the session was oversubscribed and eventually spread over two tables. This and the lively, well-moderated conversation demonstrate the level of interest in these topics.
Crowdsourced AI evals
I don’t usually choose silent disco style conference sessions, but I had to see the Legal Benchmarks presentation by Anna Guo and Elgar Weijtmans. Legal Benchmarks runs crowd-sourced evaluations of how AI models and legal AI platforms perform specific legal tasks based on feedback from 3,000 (volunteer) legal professionals in 72 countries. The findings are published online and freely accessible. While it is still early days for Legal Benchmarks, as the models and applications on their leaderboards have relatively few verified reviews, Guo and Weijtmans are producing a genuinely independent evaluation of performance and token use in a rapidly evolving and consolidating legal AI market. Most firms can’t afford to follow the example of Hogan Lovells Cadwalader which has rolled out both Harvey and Legora, so Legal Benchmarks is a useful reference for decision makers looking for best tools for their firms and in-house legal teams.
Meeting Louis Theroux
On day 2 it was important to get in early to get a place in the room for the Legal Geek celebrity interview with documentary maker Louis Theroux. It was worth the wait because Louis Theroux had plenty to say, is naturally entertaining, and is obviously comfortable being interviewed on stage. Long may this fun and interesting aspect of Legal Geek continue – although I am biased, as I like Louis Theroux’s work and I got to meet him afterwards at Thomson Reuters CoCo Lounge.
Intake trinity
Coinciding with Legal Geek, the two most prominent legal AI platforms, Harvey and Legora, launched identical agentic products for in-house legal within 24 hours of each other. Both products are called Intake, and describe themselves as a ‘front door’ for managing the requests that the legal department gets from the business, organising the work and preparing initial drafts.
While on closer inspection there are differences between how the two new Intake products approach agentic triage, when I checked the newswires on day 1 of Legal Geek, I wondered whether Alex Denne’s Hargora April Fools I wrote about six months ago had become reality. This was certainly not the case – Harvey and Legora each had a very separate (and significant) presence at Legal Geek. However, there is clearly convergence in vendor roadmaps targeting corporate legal departments. This is further evidenced by the fact that UK legal AI vendor Wordsmith had pipped them both to the post having launched its own Intake product in August. Wordsmith too had a big stand at Legal Geek, and is already a popular choice among in-house legal teams because of its conversational interface. As the legal AI market matures and it becomes clear where and how AI adds value, there is likely to be more convergence and law firms and legal departments need to identify the best option for their particular needs, which is where resources like Legal Benchmarks come into play.
Learn more about upcoming Legal Geek events on our events page.
