LegalQuants · member, mentor, Programme Director
Lawyers who build.
LegalQuants is a community of lawyers who build their own technology to practise law. I have been a member since its early months, mentored on two cohorts of its AI Residency, built two of the skills in its plugins for OpenAI's Astra for Law, and from November 2026 I am programme director for Cohort 4, assisting Jamie Tso, who runs the Residency. Everything described here is LegalQuants' work; this page sets out my part in it.
What it is
A community with an entrance exam.
LegalQuants describes a legal quant as a lawyer who builds their own tech to practise law, and it holds its members to that. The admissions floor is LQ Assess: a timed, observed session in which a lawyer builds a working tool and is graded by practising legal quants. Advanced prompting or workflow design is not enough; the test is whether you have actually built something and can explain it. Inside, members share work through channels, a directory, a shared knowledge base and roundtables, and contribute open-source work to the legal plugin ecosystems of both OpenAI and Anthropic.
What I value most is the honesty. People tell each other plainly when an approach does not work, including one that someone has spent weeks on, and the suggestion becomes a change that can be tested the next day. Everyone understands enough of both the law and the technology for those conversations to be useful. I have never worked with a group so effective.
The AI Residency
Four weeks, one tool, one mentor.
A resident arrives with a problem from their own practice and leaves with a working tool and the understanding to maintain it. Most residents had never opened a terminal before they started. The Residency is LegalQuants' programme, run by Jamie Tso; I describe it here because I mentor on it and help run Cohort 4, and applications, places and fees are handled by LegalQuants.
How it runs
Three one-hour sessions a week with a mentor matched by practice area, seniority and time zone, with the mentor reachable in between. Two recorded webinars a week, group clinics and office hours. Residents put in ten to fifteen hours a week.
How it ends
LQ Assess in the final week, then Demo Day on 6 December, where each resident presents what they built. Passing LQ Assess earns the AI Residency Graduate credential and admission to the community.
What people build
Cohort 3 produced, among other things, a computer-vision model for IP cases that beat two frontier models on its author's benchmark, a cross-examination simulator you argue with out loud, an 8-K drafting tool that exports a Word blackline, and a live tracker that replaced a spreadsheet a firm kept by hand.
Cohort 4
Runs from 1 November to 6 December 2026, with an optional Foundations week on Claude Code and Codex from 2 to 6 November for anyone starting from scratch. Fees, the application and the refund terms are on the Residency site.
Cohort 4 · November–December 2026
Schedule summary checked against the Residency site on 11 October 2026.
- 1 November
- Kickoff and intake.
- 2–6 November
- Optional Foundations: daily clinics and exercises; setup support for the cohort.
- 9–27 November
- Three build weeks with mentoring and Tuesday/Thursday webinars.
- 30 November–6 December
- Assessment, submission and review. Demo Day on 6 December.
How I mentor
The resident builds. I do not.
The temptation on a mentoring call is to take the keyboard. I try not to. The resident has to leave able to explain every part of what they built, because that is the only version of the tool they will be able to maintain, defend to a partner or extend when the next model arrives. My job is to explain the architecture when it is confusing, name the trap before they fall into it, insist on a plan and a test before anything is trusted, and say plainly when an idea is too ambitious or not ambitious enough.
In the programme director role from Cohort 4, that is the standard I will be arguing for across the mentor bench: residents who can build, test and explain their own work, and mentors who are judged on how much their resident learned rather than how much the mentor did.
"Knowing the law used to be the edge. Now the edge is knowing what to build."
"I still can't believe this metaverse was singlehandedly built by a lawyer."
For firms and teams
Sending lawyers through it.
Firms now pay for their lawyers to attend, and The Lawyer reported in October 2026 that Weil and Sheppard Mullin are among them. If you are considering it for a team, the questions worth settling first are who should go, what each of them should build, how the firm's data and tools can be used safely during the programme, and what happens to the tools afterwards. I am happy to talk those through before you apply. I do so on LegalQuants' behalf, as a mentor and, from Cohort 4, programme director: the Residency is LegalQuants' programme and is run by Jamie Tso, applications, fees and contracts are with LegalQuants, and the official routes to a place are the Residency site and a call with Jamie. My own consultancy is separate, and I will always say which hat I am wearing.
- Residencyresidency.legalquants.com
- Communitylegalquants.com
- Pluginslegalquants.com/lqplugins
- CoverageThe Lawyer, 8 October 2026
The AI Residency, LQ Assess and the LegalQuants plugins are LegalQuants' programmes and products. I describe them here as a member, a mentor and, from Cohort 4, the Residency's programme director under Jamie Tso, on LegalQuants' behalf and not as offerings of my own consultancy.