A curated guide to how we build AI-powered legal technology
July 22, 2026
Welcome to our Product Engineering Guide. It’s a curated collection of posts that tell the story of how we build at Orbital. Whether you’re considering joining us or just curious about our approach, this is the best place to start. We’re genuinely proud of what we’ve built and excited to share it with you.
How We’re Adapting to AI-First Engineering #
** Everyone Ships Code Now** — How Orbital is adapting to the age of agentic coding: tiny teams, new hiring criteria, and a future where anyone in the organization can ship to production (Andrew Thompson, CTO)
** Three Core Skills, One Team** — Why we’re rebuilding Orbital around tiny teams of 2–3 people who combine build craft, product judgement, and design taste while orchestrating fleets of agents (Andrew Thompson, CTO)
** Would You Trust an Agent to Buy Your House?** — Lessons from building an agentic lawyer, from our talk at the London Agentic AI Meetup (Éanna Morley, Senior AI Engineer)
** Grading Ourselves Against the Harness Engineering Playbook** — Scoring our own agent harness against OpenAI’s open-sourced harness-engineering playbook, twelve theses at a time (Andrew Thompson, CTO)
Scale & Impact #
We’ve processed over 100 billion tokens through our AI systems, making us one of the highest-volume legal AI applications in the world. It’s been quite a journey and here’s what it looked like:
** From Launch to 100 Billion Tokens** — A look back at 2025: the year Orbital Copilot went from ambitious experiment to industry milestone (Andrew Thompson, CTO)
** We’ve Raised $60M to Automate Real Estate Law With AI** — Announcing our Series B fundraise led by Brighton Park Capital (Andrew Thompson, CTO)
How We Build Product #
Our AI Assistant, Orbital Copilot, thinks like a real estate lawyer. Here’s a look at how we build and rigorously evaluate it:
** We Built an AI Assistant That Thinks Like a Real Estate Lawyer** — The evolution from classical ML to LLMs to AI Agents, and how we’re accelerating lawyers’ work by 70% (Andrew Thompson, CTO)
** Real World Blind Evals** — How we test new models against our production workloads before rolling them out to customers (Ben Sainsbury & Gus Levinson, Legal & AI Engineering)
Technical Deep Dives #
We love sharing what we learn. These posts explore some of the technical approaches we’ve developed along the way:
** Give Your LLM a Terminal** — How giving language models access to tools like a terminal unlocks powerful agentic capabilities (Matt Westcott, VP of AI)
** CTO’s Talk for AI Engineer World’s Fair 2025** — A conference talk on how we’ve built Orbital Copilot and the lessons learned along the way (Andrew Thompson, CTO)
** Teaching AI to Read Property Boundaries** — Building computational geometry systems that parse historical property descriptions (Andrei Vasilcoi, Tech Lead)
** No Agent Left Behind** — The engineering habits that carried our agent platform through 7× growth—graceful deploys, fair scheduling, and catching silent failures (Andrei Vasilcoi, Tech Lead)
Life at Orbital #
We’ve built a flexible, trust-based environment where people do their best work, grow their careers, and sometimes even go on to build their own companies:
** From Orbital to YC** — How working at Orbital prepared one of our AI Engineers to co-found a YC-backed startup (Gus Levinson, AI Engineer)
** What is a Legal Engineer?** — Follow a day in the life of a commercial property lawyer turned Legal Engineer, working at the intersection of law and AI (Tassneem Attia, Legal Engineer)
** Flexible Working at Orbital** — What “flexible working” actually means at our company—autonomy, trust, and putting customers first (Andrew Thompson, CTO)
** Our Product Engineering Mantras** — The principles that guide how we build: ship early, bet on the model, and more (Andrew Thompson, CTO)
Join Us #
If these posts resonate with you, we’d love to hear from you. We’re always looking for people who want to solve hard problems at the intersection of AI and law and have fun doing it. Don’t see a perfect match? Feel free to connect with me directly on LinkedIn: Andrew Thompson. I’m always happy to chat over coffee or a video call—no pressure, just a conversation.