Mike (MikeOSS) is an open-source legal AI platform for document review, drafting, and legal research. It pairs matter-centric chat with reusable review workflows, citation verification, and a document library, and it is built to be self-hosted rather than consumed as a closed service. The code is published under AGPL-3.0 in the public Open-Legal-Products/mike repository.[1][2]
Quick verdict
Teams that need legal AI inside their own infrastructure benefit most from Mike: its Docker Compose deployment, evidence-grounded answers, and CourtListener-backed research keep client documents under the deploying team's control. The constraint is responsibility: the team supplies its own model key or runs local models, operates the stack, and accepts a beta Word add-in. Teams wanting self-hosted legal AI that accept running the infrastructure should choose Mike; teams wanting a managed product with no operations burden should choose a closed legal AI service.[1][2]
Key features
Matter-centric document chat Matter-centric document chat: ask questions of legal documents and open matters with evidence-grounded answers tied to the underlying files.[1][2]
Document review with suggested edits Document review with suggested edits: run reviews that propose concrete edits to apply rather than returning abstract commentary.[1]
Reusable assistant workflows Reusable assistant workflows: package repeatable review processes as maintained assistant and tabular-review workflows instead of one-off prompts.[1]
Projects, folders, and library organization Projects, folders, and library organization: keep matters, documents, and templates structured the way a practice already files work.[1]
Citation verification and case-law research Citation verification and case-law research: check citations and research US case law through the CourtListener integration.[1][2]
Beta Microsoft Word add-in Beta Microsoft Word add-in: work from Word through a task-pane add-in while it remains in beta rather than assuming full editor parity.[1]
Local models through Ollama Local models through Ollama: run supported language models locally instead of sending matter content to a vendor model API.[1]
Connectors and tamper-evident exports Connectors and tamper-evident exports: connect Google Drive, Gmail, and Calendar with explicit approval for sends and edits, link MCP connectors, and produce tamper-evident exports for defensible handoffs.[1]
Who will love it
Firms that must self-host: the Docker Compose stack and deployment guide keep client documents on infrastructure the team controls.[1][2]
Teams with repeatable review work: reusable assistant and tabular-review workflows turn recurring document reviews into defined processes.[1]
Practices sensitive to model-data exposure: Ollama local-model support plus evidence-grounded answers reduce how much matter content must leave the firm's systems.[1][2]
Who should avoid it
Teams wanting zero operations: running Supabase, storage, secrets, and model keys is real infrastructure work with no documented managed offering.[1]
Lawyers living entirely in Word: the Word add-in is explicitly beta, so Word-first drafters should not assume it matches the web workspace.[1]
Buyers needing a supported product contract: no enterprise support offering was found on the inspected official pages, so teams that require SLAs and vendor accountability will miss them.[1]
Alternatives and compatibility
No legal product exists in this catalogue yet, so Mike has no supported catalogue relationship in this pass and this comparison is editorial context, not a sourced edge.[1][2]
Closed legal AI services fit teams that want a managed product with no infrastructure to run.[1]
Documents and matters live in the deploying team's Postgres and object storage, so switching later is a data-migration project owned by the team, not an export button.[1]
Questions worth asking
What is the best Mike alternative for non-technical users?
There is no non-technical alternative inside this catalogue yet, since Mike is its first legal product. For non-technical buyers the honest answer is a managed closed legal AI service, because Mike's value assumes a team that can deploy and operate a Docker Compose stack with secrets and model keys.[1][2]
Is Mike really no-code?
Yes for its purpose: chatting with matters, running reviews, and organizing the library needs no scripts or configuration from the lawyer using it. The technical part is operational, since deploying Mike means running containers, storage, and model access.[1]
How much does Mike cost per month?
Nothing for the platform itself: the AGPL-3.0 core has no paid edition documented. The real monthly costs are the operator's own infrastructure and model usage, since teams add their own Anthropic, Gemini, or OpenAI key unless they use Ollama exclusively. Checked October 2026.[1]
Can you self-host Mike?
Yes. The README provides a Docker Compose stack running Mike with Supabase and R2-compatible object storage, plus a manual and production deployment guide. Bundled credentials are for local development only, so production use means supplying real secrets, storage, and model access.[1]
Sources and editorial review
Source research. Reviewed on October 11, 2026. Written by Sanjib Sah. AI assisted; reviewed by a human editor.
We have not run or tested this product.
- Mike homepage (official)
official · Observed October 11, 2026
Official homepage: self-hosted open-source positioning, matter chat, review, drafting, citation checking, evidence-grounded answers.
- Open-Legal-Products/mike README (official)
official · Observed October 11, 2026
Official repository README: stack, Docker Compose, Ollama, CourtListener, Word add-in, connectors, licensing.
Compared with these products
No explicit proprietary baseline is recorded.