Comp AI
Open-source, agentic-first CRM with a built-in research agent that books follow-ups, enriches contacts, and keeps your pipeline up to date.
Comp AI is an open-source, agentic-first CRM built by the team behind trycomp.ai. Most CRMs are a database with a form in front of it; the AI ones bolt a chat box onto the side of that form. Comp AI is built the other way round — the agent is not a feature of the CRM, the CRM is where the agent keeps its notes. It runs on its own deployment, on its own schedule, against its own work queue, deciding what to look at next, booking its own follow-ups, and stopping when the research budget runs out.
The stack is a TypeScript Turborepo monorepo on Bun: a Next.js front end, a NestJS API with tRPC type safety, Prisma on Postgres, and a research agent built on the eve durable-agent framework. The agent ships with 18 tools (read CRM history, search, identify contact, research person, enrich company, record fact, schedule recheck) and 4 skills, runs inside a network-isolated sandbox, and is provider-agnostic via the Vercel AI Gateway.
Comp AI is single-tenant and self-hosted by design. Google sign-in is the only door; an allow-list env var is the entire authorisation model. Mailbox and calendar sync keep deals, contacts, and companies current without a human typing them in, and every outside data source (LinkedIn, web research, company brand data) is optional — the agent works with none of them. It is released under the MIT license and targets internal sales teams that want a self-hostable Salesforce / HubSpot / Pipedrive / Zoho CRM alternative without an AI feature bolted on as an afterthought.
