Straight answers about Odigos -- what it is, how it works, what it costs, and how to install it.
A self-hosted personal AI agent with persistent memory, self-improving behavior, and autonomous sub-agents. You deploy it on your own hardware (Docker or bare-metal Python 3.12+) and connect any OpenAI-compatible LLM provider. Your data stays on your machine -- no cloud dependency, no shared backend.
Self-hosted is always free -- MIT license, one bash install.sh away.
Managed hosting is $15/month: LLM included, all features enabled, isolated container, zero setup. See full pricing.
Additional tiers -- Pro and BYOK -- are in development.
Python 3.12 with FastAPI and aiosqlite on the backend; React 19 + Vite + Tailwind + shadcn/ui on the frontend. sqlite-vec for vector search, FTS5 for keyword search, cross-encoder reranking on top. One process, one SQLite database, no microservices.
Two paths:
git clone https://github.com/tamler/odigos && cd odigos && bash install.shbash install-bare.shSanity-check the install with bash tests/e2e/test-cards-e2e.sh. Typical hardware: 1 CPU and 1 GB RAM per agent -- a $40/month VPS comfortably hosts 30 to 50 agents side-by-side.
Yes. Self-hosted users have full control -- your agent's memory, database, and API traffic never leave your hardware. On managed hosting, each agent runs in its own isolated container with its own database. See the Privacy Policy for details.
Your data is yours. Export your knowledge graph anytime. Your agent stops running at the end of the billing period; data is retained for 30 days for recovery, then deleted.
Yes. Bring your data with you in either direction.
For billing or account issues, reach us via the contact form. For code-related questions, feature requests, or bug reports, open an issue on the GitHub repository.