The archive · AI & Models · Product decision · 2026
Open Brain bets every AI tool can share one persistent memory; GitHub hits 4.6k stars
Open Brain (OB1) is a self-hosted memory database and AI gateway so Claude, ChatGPT and Cursor share one brain; GitHub 4.6k stars by 2026.
Open Brain
What the business is
Open Brain (OB1) is an open-source personal-AI infrastructure layer: a vector-search database of thoughts plus an AI gateway and a chat channel (Slack or Discord), exposed to Claude, ChatGPT, Cursor and other agents through MCP so every tool reads and writes the same persistent memory — explicitly no middleware or SaaS chains.
How it started
Nate B. Jones created Open Brain in March 2026 (repo created 2026-03-11) as open-source infrastructure rather than a notes app: 'a database with vector search and an open protocol — built so that every AI tool you use shares the same persistent memory of you.' The README's promise is one database, one AI gateway and one chat channel, with any AI able to plug in and no middleware, SaaS chains or Zapier.
What happened
Jones built a small maintainer team (repo manager, community admin, community maintainer), opened contributions under an FSL-1.1-MIT license, and set up an automated review agent that checks structure, secrets, SQL safety, dependencies and documentation before human review. The community contributed import recipes for ChatGPT, Perplexity, Obsidian, X, Instagram, Gmail and Google Takeout exports, plus skills, dashboards and integrations such as a self-hosted Kubernetes deployment with PostgreSQL and pgvector, an OB1 Agent Memory runtime for OpenClaw, and a Next.js dashboard.
How it ended up
Still running and growing as an open-source project as of 2026-09-05: 4.6k stars, 879 forks and 716 commits, with no funding, revenue or incorporation disclosures in the record.
Background
Open Brain (OB1) is an open-source system created by Nate B. Jones and launched on GitHub on 2026-03-11. Its premise is that the personal AI stack is backwards: instead of every tool — Claude, ChatGPT, Cursor, whatever ships next month — keeping its own memory of you, there should be one database with vector search and an open protocol, one AI gateway and one chat channel, so every AI reads and writes the same persistent brain.
The bet is anti-SaaS: 'no middleware, no SaaS chains, no Zapier.' A user builds the full system — database, AI gateway, Slack capture and MCP server — in about 45 minutes without coding experience, then extends it. Extensions compound: home facts, scheduling, meal plans, contact tracking and an interview pipeline all read the same thoughts table, and recipes import years of exports from ChatGPT, Perplexity, Obsidian, X, Instagram, Gmail and Google Takeout.
Jones recruited a maintainer team and published a contribution pipeline: automated review agents check every PR for structure, secrets, SQL safety, dependencies and documentation, then a human maintainer reviews. That governance plus the MCP integrations — Kubernetes/PostgreSQL deployment, an OB1 Agent Memory runtime for OpenClaw, Next.js and SvelteKit dashboards — converted early users into contributors, with the repo reaching 4.6k stars, 879 forks and 716 commits by the 2026-09-04 crawl.
What has to be true
- Concrete platform evidence: the GitHub crawl of 2026-09-04 recorded 4.6k stars, 879 forks, 716 commits and 110 watchers for a repo created 2026-03-11.
- A sharp anti-SaaS bet: one self-hosted database, one AI gateway, one chat channel — 'no middleware, no SaaS chains, no Zapier' — instead of another AI notes app.
- Visible governance as strategy: every PR passes an automated review agent before a human maintainer, and merged community work is listed by name in the README.
- The wedge is capture-first: a Slack/Discord bot feeds a vector-searchable thoughts store that MCP then exposes to any AI client, so adoption starts with a 45-minute build.
What can be applied
When the moat is your data, the product is an open protocol: shared self-hosted memory beats another notes app only if every model can plug into the same store and the review pipeline stays visible.
Aftermath
As of 2026-09-05 Open Brain is live and growing as a creator-led open-source project. The repository recorded 4.6k stars, 879 forks, 716 commits and 110 watchers at the 2026-09-04 crawl, with a Discord community, a companion Substack, and a community contribution system that merges recipes, skills, dashboards and MCP integrations after automated and human review. The record shows no funding, revenue, pricing or incorporation disclosures; monetization appears limited to the author's Substack and companion materials, so the bet is still in its adoption phase.
Sources
spotted an error? The archive wants to know.
Your turn
You just read one. Describe what you are building, and see who is betting on the same thing.
Free account · 3 free questions · no card