Our Take
Mike proves legal AI can be built outside vendor walls, but Chen's 'viable alternative' claim needs independent validation beyond LinkedIn testimonials.
Why it matters
Small and mid-size firms priced out of Harvey's enterprise deals now have a self-hosted option that keeps documents internal. BigLaw firms get leverage in vendor negotiations.
Do this week
Legal tech teams: Test Mike's document review capabilities against your Harvey workflows before year-end budget cycles close.
Former Latham lawyer builds Harvey competitor in open source
Will Chen, a former Latham & Watkins attorney, released Mike, an open source legal AI platform that replicates Harvey's core functions. The platform includes document review, creation and editing tools, project management (Harvey's "Vault" equivalent), tabular review, and workflow automation.
Mike runs locally on firm infrastructure rather than cloud-hosted servers. Law firms download the code and deploy it on their own systems, keeping client files internal. The hosted version at mikeoss.com serves as a demo only.
The project gained 1000 GitHub stars and 300 forks within 72 hours of launch (per Chen's interview). Chen reports that small and medium-sized firms have successfully deployed local versions, with positive feedback posted on LinkedIn and X.
Security concerns drive demand for self-hosted legal AI
Law firms have hesitated to adopt Harvey's newer features like direct agent connections to document management systems. Some firms prohibit lawyers from uploading confidential documents to Harvey's cloud-based Vault due to security policies.
Mike addresses these concerns by eliminating third-party data storage. When firms control the entire software stack, client files never leave their infrastructure. This closed-loop approach resolves many law firm security objections to cloud-based legal AI.
The project also challenges legal AI pricing models. While BigLaw firms can afford Harvey's enterprise contracts, smaller firms often get priced out. Mike provides these firms with comparable functionality at the cost of LLM tokens and internal deployment.
Early validation, but scale questions remain
Chen acknowledges that Mike works for small and medium firms but needs additional development for large firm and enterprise deployment. The platform can render complex documents like CP checklists properly and make tracked changes across multiple documents simultaneously within projects.
However, independent benchmarking remains limited to user testimonials on social media. Chen positions this as "early public validation" but provides no comparative performance data against Harvey or other legal AI platforms.
The project uses AGPL v3 licensing, allowing firms to modify code for internal use while requiring source code sharing if they provide third-party access. Chen is considering switching to more permissive open source licenses.