Happy New Year Friends 👋
In our season two opener, we sit down with Anson Lai - commercial counsel by day, relentless tinkerer by night, who walks us through exactly how he built, refined, and published a document review tool as a Microsoft Word add-in that rivals alternatives from legal AI startups raising hundreds of millions of dollars.
The kicker? He did it in weeks, not years. And he’s giving it away.
Listen Now
Available to watch or listen on Spotify, Apple Podcasts, and wherever you enjoy your podcasts.
This isn’t theoretical. In this highly insightful exchange, Anson shares his screen (so watch if you can, rather than just listen), shows us the tool live, explains the design choices (”Keep It Simple Stupid” is apparently both a philosophy and a technical strategy), and crucially opens the hood on what actually makes it work. No mystique. No black box. Just a lawyer who got tired of copy-pasting contracts into ChatGPT tabs and decided to do something about it.
What You Might Learn
The “Vibe Coding” Approach — Anson describes how talking to AI tools (rather than just instructing them) shaped better technical decisions. The conversation he had with his IDE mattered as much as the code that came out of it.
Why “Bring Your Own Key” Changes Everything — No servers sitting in the middle. No data passing through third parties (including Anson). Your document goes straight to Google’s API and comes straight back. Anson walks through exactly why this architecture matters.
The Real Competitive Moat — If building end to end software products now takes hours (not months), then the differentiation isn’t likely to lie in creating the thing. Anson suggests it all now lies in the refinements and iterations, admitting he’s been polishing edge cases for weeks (think nested numbered lists, tables, sub-sub-sub-sections) because that’s how real documents evolve and where most gen AI tools quietly fall apart.
Students, Cover Letters, and Beyond — This wasn’t built just for lawyers. Anson designed it with students in mind: but more broadly anyone who writes, reads, revises and occasionally stares at a blank page on a Friday afternoon needing a nudge.
The implications
We’re genuinely curious to see how 2026 plays out.
It isn’t unconscionable to perceive a growing community of solo builders, small in-house teams and law firms quietly building and deploying tools to production that do 80% of what the big players offer - at a fraction of the cost, with full transparency, and tailored to the exact workflows that matter to them.
Anson’s Word add-in is one example.
Jamie Tso (who we featured last season) is building similar things.
They both form part of an emerging cohort of legal technologists who aren’t waiting for permission or procurement cycles - they’re just building.
It’s going to be fascinating to watch.
Key References from Our Conversation
Google Antigravity Google’s AI-powered IDE announced in November 2025, built as a fork of VS Code. Supports multiple AI models including Gemini 3 and Claude Opus 4.5. Anson used this environment to build his add-in through conversational prompting—a textbook example of “vibe coding.”
Project IDX / Firebase Studio Google’s earlier cloud-based IDE, now transitioned into Firebase Studio. Referenced as an alternative development environment for building web and multi-platform applications.
Gemini 2.5 Flash / Gemini 3 Flash / Gemini 3 Pro Google’s family of AI models. Anson primarily uses Gemini Flash for speed and cost-efficiency, with Pro available for more complex reasoning tasks. Gemini 3 Flash was released in December 2025 and outperforms previous Pro models while being 3x faster.
ElevenLabs AI text-to-speech platform that Anson mentions using to convert articles, blog posts, and even contracts into audio for listening during jogs. Part of his personal workflow for consuming vast amounts of written content.
Connect with Anson
LinkedIn: linkedin.com/in/anson-lai
Gemini AI for Office – Microsoft Word Add-In Anson’s open-source Word add-in available on GitHub. A bring-your-own-key solution that provides chatbot functionality, redlining, commenting, highlighting, Google search integration, and playbook features—all running directly against Google’s Gemini API with no intermediary servers.
Law://WhatsNext is back for Season 2.
A reminder that we’re all about exploring how leading practitioners are leveraging emerging technologies and ways of working to pursue their passion and objectives, and as a by-product we’re curious to understand the implications for the future of legal practice?
No hype. No hand-waving. Just practical conversations with lawyers, educators and technologists trying to leave their mark on the future!
Welcome back. Let’s get into it.
Tom & Alex
