A modern, sovereign AI backend for products
and enterprise deployments.
Run it · Try it · Documentation · Architecture · Contributing ·
Get the breadth of a managed AI platform with control over the infrastructure, data and extension points. Use it behind customer products, internal systems, and regulated deployments without handing the AI control plane to a managed provider.
- Agent builder and runtime
- Multi-provider model gateway
- Knowledge ingestion and retrieval
- More than 200 typed integrations
- MCP, OpenAPI, GraphQL and code tools
- Sandboxed code and shell execution
- Web widgets, portals and messaging channels
- REST and GraphQL APIs, webhooks and generated client types
- Node.js, Python and Go SDKs and a Terraform provider
- Authentication, users, teams, contacts and multi-tenant identity
- Access control, moderation, PII protection and audit
- Traces, events, ratings, usage and operational logs
- Replaceable database, storage, cache, queue and vector infrastructure
Run the complete prebuilt stack with one command - no checkout, no build:
docker compose -f oci://ghcr.io/chatbotkit/platform-community:latest upOpen http://localhost:3000. Sign in with any email address and read the six-digit code from the platform container log:
docker compose -f oci://ghcr.io/chatbotkit/platform-community:latest logs platformSee Deployment for details.
Binary assets are stored with Git LFS, so install it (git lfs install)
before cloning. From a fresh checkout, run
docker compose upOpen http://127.0.0.1:8080. See Getting started for host-side development, storage configuration and the first model connection.