// the find
mindsdb/minds
General-purpose AI designed for knowledge workers — creators, strategists, and operators — and individuals seeking AI systems they can truly control to help them get work done, with full flexibility to extend and deploy anywhere (VPC, on-prem, or cloud).
MindsDB's latest pivot: a general-purpose AI workspace for knowledge workers, rebranded from its original SQL-interface-for-ML roots. The actual implementation lives entirely in git submodules (core_agent, core_api, data-vault, frontend) — this repo is a thin orchestration shell with a Makefile and docker-compose. The 39k stars are inherited reputation from the original MindsDB product, not this one.
- The deployment story is genuinely useful for enterprise: VPC, on-prem, air-gapped, and cloud all supported, which puts it ahead of most hosted-only AI tools for regulated environments.
- MCP support in the topics means it plugs into the emerging Model Context Protocol ecosystem, which is where agent tooling is converging.
- The original MindsDB data connector heritage — BigQuery, MySQL, PostgreSQL, MSSQL, and more — gives this a real leg up on data access compared to starting from scratch.
- Active development with a large contributor base and a proper security disclosure policy — not a weekend side project.
- All substantive code is in opaque git submodules (core_agent, core_api, data-vault, frontend are commit hashes with no inline code). You cannot audit, fork, or contribute to what actually runs without chasing down four separate repos.
- The README is almost entirely marketing copy. There is no technical architecture description, no API surface, no explanation of what 'Minds' actually does differently from running your own LLM + RAG stack.
- The star count is misleading — this repo is riding the reputation of the original MindsDB SQL-for-ML product, which was a genuinely novel idea. This rebrand into 'AI for knowledge workers' puts it in a crowded market (Notion AI, Microsoft Copilot, etc.) with no clear technical differentiation stated.
- Primary language is Dockerfile. When the only thing in your repo is config files and a Makefile, that is a sign you are looking at a product wrapper, not a project you can meaningfully self-host or extend without significant undocumented work.