An ex-Big Law associate built a legal AI platform in two weeks - and it's going viral

A former Latham associate has taken a big swing at the legal AI market - by building his own open-source tool he says does everything the major legal AI platforms can, without the monthly fee.

In a viral social media post, Will Chen unveiled Mike OSS - or just “Mike” - a vibe-coded tool built in two weeks which he claims has feature parity with platforms like Harvey and Legora.

The product includes an AI assistant for drafting and reviewing documents, a file-sharing vault, bulk document review and pre-built workflows.

Build it yourself

Chen began learning to code before the generative AI revolution as a law student at Oxford. While a trainee at Latham & Watkins, he created Lawprof.co, an online learning platform for law students.

Speaking to Non-Billable, Chen said he thinks lawyers overestimate how hard legal AI is. He said many tell him they are using tools for tasks that general-purpose models such as ChatGPT or Claude can already handle - pointing to summarising documents, proofreading and drafting emails as examples.

Mike, vibe-coded with Claude, is free to use, available to firms as a hosted product or to self-deploy from source.

“The product is meant to be a proof of concept but it is also fully functional,” Chen said, adding that lawyers and tech people had already reached out to say they had set up their own local versions, “and it is working great for them.”

He also claims the system can provide citations to relevant passages when reviewing documents, and has additional version control functionality and an IDE-like interface for assistants - although it lacks features such as Word or Outlook integrations.

In defence of the vibe-code approach, Chen said: “really every single software company in the world, from Harvey and Legora, to tech giants like Meta, OpenAI and Anthropic are vibe-coding right now.

Vibe-coding is starting to become the default method of software programming. Vibe-code does not always imply slop code. It just depends on how you use it.”

Could it scale?

Among the hundreds of comments online, several pointed out that enterprise-ready software is about more than core functionality - particularly in legal, where buyers care most about things like data security and client confidentiality.

On the question of security, Chen told us Mike is designed to be run inside a law firm’s own systems, so documents and data never leave their control. He said that while some enterprise features are still being built, the open-source approach means security can improve quickly over time.

Chen joins a growing wave of lawyers turning to AI to build their own products. One of the more prominent examples is Jamie Tso, who recently left his job as a funds lawyer at Clifford Chance to pursue legal AI full-time, working with in-house teams to build custom tools.

For now, Mike feels more like a provocative proof of concept than a direct competitor to the incumbents - but it taps into an interesting question for the market: how much of legal AI is the product itself, and how much is everything around it?

Any similarities between Mike OSS and persons fictional or real - including those from a well-known legal drama television series - are merely coincidental, according to its founder.