Best Practice

Andrew Thompson on What Legal AI Can Learn From Software Engineering

Guest: Andrew Thompson Role: CTO, Orbital LinkedIn: https://www.linkedin.com/in/andrewmatthewthompson/ [https://www.linkedin.com/in/andrewmatthewthompson/] Website: www.orbital.tech [http://www.orbital.tech] (company) | www.andrewthompson.co [http://www.andrewthompson.co] (personal blog) This AI platform powered over 200,000 property transactions last year and raised $60M in January to expand even further. I spoke with Andrew Thompson, CTO at Orbital. Andrew previously built the Caffè Nero app, ran product and tech at Appear Here, founded his own AI company, and now leads engineering, product and a team of in-house real estate lawyers at Orbital. We discussed: → How Orbital mix both generalist LLM’s and their own classical ML into one product. → Why legal is roughly 12-18 months behind software engineering, and how to use that lag. → What “ship early and often” actually looks like inside a regulated industry. → Andrew’s tips and advice for Andrew is an absolute genius in everything LLM’s and how to make the most of it in the physical space around us. I really hope you take away something useful from this. Want to feature in the Best Practice podcast? Email george@georgehannah.com [george@georgehannah.com] Sources mentioned in the episode: * Orbital’s $60M Series B (Jan 2026, led by Brighton Park Capital): https://tech.orbitalwitness.com/posts/2026-01-23-series-b-announcement/ [https://tech.orbitalwitness.com/posts/2026-01-23-series-b-announcement/] * Caffè Nero (Andrew’s previous build): https://caffenero.com * Appear Here (Andrew’s previous build): https://www.appearhere.co.uk * Seedcamp (where Andrew was a startup mentor): https://seedcamp.com * ElevenLabs (referenced re: agentic insurance): https://elevenlabs.io * Legalweek New York 2026 (referenced throughout) * McKinsey research on automatable legal work (referenced) * OpenAI ChatGPT and GPT-4 (referenced as the inflection point) * Anthropic Claude (referenced) * Google BERT and T5 (Orbital’s pre-LLM stack) This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit bestpracticeai.substack.com [https://bestpracticeai.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

Meet The Startup Building AI For The Messiest Part of Law

“Crack the chronology, crack the case.” That’s what Gregory Mostyn‘s [https://www.linkedin.com/in/gregory-mostyn-9b3a89b8/] father - a barrister, then judge for 42 years - was told on his first day in chambers. Decades later, it’s the founding insight behind Wexler. I spoke with Gregory Mostyn, co-founder and CEO of Wexler, the fact intelligence platform now used by huge massive circle firms in their litigation teams. Greg isn’t a lawyer. He grew up watching his father - who handled some of the highest-profile family cases in England - come home with ten ring binders to read before court the next morning. That image, plus a stint at Entrepreneur First, became Wexler. The company has 20x’d ARR since pre-seed and just raised $5.3m led by Pear VC. We discussed: → Why Wexler went deep on litigation when every other legal AI startup was going broad → The “verification tax” and where fully agentic legal AI actually breaks down → Why consumption pricing beats per-seat for litigation work (and how they structure it) → How his 67-year-old father is now vibe-coding apps for his tennis club → What it takes to close a Big Law deal in a week I absolutely loved recording this episode and I hope you enjoy listening to it just as much as I did. Know someone that might want to come on the Best Practice Podcast? Email me at george@georgehannah.com This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit bestpracticeai.substack.com [https://bestpracticeai.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

Meet the Startup Building the Matter Management Layer Legal AI Forgot About

I keep hearing about legal AI tools that do the work better. What I haven’t heard, until now, is anyone building the tool that remembers what the work is actually for. In this episode I chat with Dillon Hirandiran, founder and CEO of Counsel (gc.inc) Dillon previously worked in investment banking, then built his first startup during Covid. The experience of navigating legal work as a founder - seed rounds, Delaware flips, cap tables - is what first drew him to the problem Counsel is now solving. Counsel is an autonomous matter management system for law firms - reading every email, every message, and maintaining a persistent context layer across everything a firm has going on. We discussed: → Why the hardest problem in legal AI has never really been document generation, but instead it’s CONTEXT retention across matters. → Why clients can’t always evaluate the quality of legal work, only its responsiveness → What a “10x attorney” actually looks like when AI handles the document layer, and why being a workhorse is the worst thing a junior lawyer can be right now → Why the biggest risk to Big Law isn’t pricing pressure from in-house or AI, but is actually their own talent leaving. Want to sponsor or feature on the Best Practice Podcast? Email george@georgehannah.com This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit bestpracticeai.substack.com [https://bestpracticeai.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

How this ex Amazon and Replit lawyer built the Legal AI platform she always wanted.

Cecilia Ziniti’s started as a paralegal at Yahoo in the early 2000s, when Yahoo was competing with Google and the internet was still being shaped by questions about content moderation, trust and safety, and what it meant to serve millions of users. She went to law school, then joined Morrison & Foerster, where she represented big brands like Apple. At Amazon, she was the founding lawyer on Alexa - she remembers getting excited about the first user review for a product that would eventually ship billions of devices. She eventually became general counsel at Replit, the developer platform used by millions of programmers. It was at Replit, in early 2022, that things changed. Cecilia had access to a pre-ChatGPT version of GPT through a deal she herself had negotiated between Replit and OpenAI. She tried it on a legal question. The result gave her chills. Her boss at the time, Replit CEO Amjad Masad, eventually noticed her attention had shifted. He asked whether her head was still in it. It was not. She left Replit and founded GC AI a week later, alongside co-founder and CTO Bardia Pourvakil, a fellow Replit alumnus. The company has since raised $73 million in total funding, including a $60 million Series B led by Scale Venture Partners and Northzone. It now powers over 1,400 companies, from Zscaler to Liquid Death. Cecilia was recently named to Inc.’s 2026 Female Founders 500 list. Would you, or anyone that you know like to feature on the Best Practise podcast? Email: george@georgehannah.com This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit bestpracticeai.substack.com [https://bestpracticeai.substack.com?utm_medium=podcast&utm_campaign=CTA_1]

How this YC-backed startup are reinventing the law firm from the ground up

I spoke with Javed Qadrud-Din, co-founder and CTO of General Legal - the AI-native law firm backed by Y Combinator, built to serve growth-stage companies. Javed has been coding since the age of nine. He went to Harvard, detoured into law at Fenwick & West, then decided he’d rather build things than advise on them. He got deep into deep learning in 2014, built what’s been described as the first semantic search system in legal at CaseText - before GPT existed - and was part of the team behind a multi-million exit to Thomson Reuters in 2023. Now at General Legal, Javed is rebuilding how legal services are delivered: flat fees, and a three-hour turnaround time. We discussed: → Javed’s extraordinary background and how General Legal came about → The difference between being AI-native and being AI-powered → What the economics of flat-fee legal services look like under pressure, and whether $500 per contract goes down as AI improves. → Why founders using free ChatGPT to draft contracts are “yoloing it”. → His vision for what the legal industry looks like in 2030 - and what lawyers should be doing today to prepare. I hope you enjoy! Want to feature on the Best Practice Podcast? Contact george@georgehannah.com This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit bestpracticeai.substack.com [https://bestpracticeai.substack.com?utm_medium=podcast&utm_campaign=CTA_1]