Mike, the Open Source Legal AI Platform – Will Chen Interview

Since former Latham & Watkins lawyer Will Chen launched open source legal AI platform, Mike, social media posts on the subject have surged across the internet. Views range from huge support for what is pitched as an alternative to Harvey and Legora, to more measured takes.  

Whatever one’s view, one cannot ignore the fact that Chen and his creation Mike have joined – and turbo-charged – a debate on an essential question: where does the value reside in legal AI tools?

Artificial Lawyer interviewed Chen and asked a range of questions. There is also a short demo video of Mike, which AL presents below.

—

First, the question everyone wants to know the answer to: Is Mike ‘as good as’ Hargora in your view? (At least in terms of what’s been built so far?)

It has the core functions of Hargora’s web application: an assistant that can help lawyers review, create and edit documents, projects (or the Vault as Harvey calls it), tabular review, and workflows. You can also chat with tabular review results like in Hargora and spin up assistants within projects.

Mike is functional, usable end-to-end for key legal workflows and I welcome everyone to test it out. It can also do some things better, such as properly rendering documents like CP checklists in landscape, and making precise edits to existing documents in tracked changes while versioning it up. While both companies’ word add-ins are able to do so, their in-browser assistants struggle a little. The assistant is also able to replicate documents and make tracked changes edits in multiple documents all at once within a project.

Is this a ‘ready to use’ alternative for real client work, when considering security, accuracy, and reliability criteria? (Which will matter a lot to all lawyers.)

A small or medium sized firm can use and get it to work. More work has to be done to increase scaleability to large firms and enterprises.

Since the launch of Mike, lawyers have successfully spun up local versions of Mike and they have reported that they were satisfied with its performance. There are quite a few public posts on LinkedIn and X. There is early public validation that it works in practice.

In terms of security, I believe that Mike offers a secure alternative to the hosted solutions that Hargora provide. To clarify, the web app hosted on mikeoss.com is a demo. The vision is for firms to take Mike’s code and implement on their own, locally or within the intranet. When they do that, their files never have to leave their computers and databases. They do not have the risk that comes with relying on a third party vendor who stores their files and workflows. Law firms take security very seriously. And indeed law firms have been hesitant in adopting Hargora’s new features like direct agent harnesses to their DMS. Some do not allow their lawyers to upload confidential documents to the Vault. But if the law firms own the software themselves, it creates a closed loop without a third party vendor and a lot of their concerns about security are assuaged.

Overall, what does Mike prove?

That legal software should and can be open and accessible, rather than high priced and exclusive, only affordable to elite firms.

That AI has fundamentally changed the speed of software development, and that perhaps we should fundamentally rethink how software, including legal software, is delivered.

I think so far the response has shown that there are many out there who share the same sentiments. On GitHub, Mike had more than 1000 stars and over 300 forks 72 hours after launch, the most for any legal tech project in history.

Why did you make this and who is it primarily for?

To challenge how legal AI is priced and delivered and to show there’s a viable alternative to expensive, closed systems. Open source tools are very commonly adopted, securely and safely, in the software industry. The legal industry can follow.

Mike is for anyone who is priced out by Hargora. It is also for those who want to fully own their AI software stack and their documents. Biglaw firms have no problem with paying for Hargora. But many small and medium-sized law firms are priced out.

How did you get into legal tech coding?

I have been coding as a hobby since university. I studied law and worked as a lawyer in Latham & Watkins for three years. Naturally, I fell into legal tech coding as it was an intersection of my personal interest and profession.

The product is free, aside from any token / LLM costs. Why not commercialise this?

I have no plans for commercialisation at the moment. My immediate goal was just to start a conversation. I also enjoy the process of building.

Also, beyond Hargora, there are already many smaller competitors offering a similar product. I think more value can be brought by making the project open source.

Have Harvey or Legora made any comment to you about this?

No.

If a law firm uses Mike, who controls how they use it? If they modify it, can they keep those changes to themselves?

They control it. They can modify the code however they want. They can also keep those changes to themselves. That is provided that they use it purely internally within the firm, which I imagine is how Mike will be implemented in most situations. But if they allow third parties to remotely interact with their version of Mike (e.g. through a client portal), they have to offer access to the source code to the third party under the AGPL v3 copyleft licence.

I am also currently considering switching to more permissive open source licences.

Many vibe-coders / legal quants seem to view what they do as an alternative to the products made by commercial legal tech businesses, but is there a way you can work together and become allies?

Yes, certainly. Legal vibe-coders play an important role as lawyers know best what product features they want. They can help to start conversations on what products should be built for lawyers. I think the question is whether all these creative people can be organised together to work on something effectively, especially outside of a formal corporate structure.

And more broadly, the field of legal tech is going through profound changes, e.g. Claude for Word, MS Legal Agent, the rise of vibe-coders, huge investments and more. Where do you see this all going?

I think thin wrappers that do not provide sufficient value are under the threat of being replaced by equivalent services from MS, Anthropic, OpenAI etc who have an immense distribution advantage. Thick wrappers with a unique value proposition and high firm-wide utilisation will survive.

Thank you, Will.

—

The debate about where legal AI product value resides will continue, especially as Claude for Word, MS Legal Agent, and a swell of vibe-coded products hit the market. AL would be glad to hear your views on this subject.

If you would like to learn more about Mike, then please see here.

—

A Legal Tech Conference For All of Europe – Legal Innovators Europe – Paris – June 24 and 25.

Express route to your ticket here.

And, 

Express route to your Legal Innovators California June 10th and 11th ticket here.

Legal Innovators California, the landmark West Coast legal tech event, will take place on June 10 and 11, in the heart of the Bay Area, the home to many of the world’s leading AI businesses – and plenty of legal tech pioneers as well! More information and tickets here.