Ten focused AI modules — customer service, knowledge, workflows, meetings, marketing and more — under one identity, one privacy posture, one open codebase.
Free to self-host. Sovereign by design. Built in the open.
Small businesses have stopped experimenting with AI — they're buying. The typical SME now runs a handful of disconnected AI tools, each with its own login, its own bill, and its own opaque cloud pipeline sending your data somewhere you can't see.
We think that's backwards. ODW·AI is one open engine with ten focused modules that share the same authentication, orchestration, and privacy boundary — so they compose into multi-step agents instead of piling up as yet another SaaS subscription.
And because the whole stack is open source, you can run it on your own infrastructure, inspect every line, extend anything, and keep every byte of your data inside your boundary.
The moat isn't the model. It's where the data lives — and who controls it.
Five modules are live and open source today. The rest arrive as we build them in the open.
| Module | What it does | Stack |
|---|---|---|
| ODW Desk | Self-hosted, WhatsApp-first AI customer support agent with full data sovereignty — PII detection, RAG-powered responses, human handoff, GDPR compliance. | Python · MIT |
| ODW Vault | Sovereign, self-hosted RAG platform with a pre-flight pipeline, hybrid retrieval, and full offline capability. | Python |
| ODW Loop | Agentic workflow orchestration engine — Fastify backend, React Flow canvas, and a DAG execution engine that lets modules compose into multi-step agents. | TypeScript |
| ODW Recap | Sovereign, on-device meeting intelligence — local transcription, AI summarization, and team sync. | Rust |
| ODW Pulse | Multilingual content & marketing generation with cultural adaptation, A/B experimentation, and infrastructure sovereignty. | Python |
Voice (AI receptionist) · Hunt (AI SDR, human-in-the-loop) · Books (AI bookkeeping) · People (AI hiring) · Shield (SMB cybersecurity)
Free to self-host. Clone it, run it on your own servers, keep your data in your boundary. No license fee, no per-seat tax, no usage meters.
Inspect every line. No black boxes. Audit how your AI handles customer data, see exactly which models run, extend anything you need.
Bring your own models. Open models (Llama, Mistral, Qwen) run locally, with optional cloud burst when you decide it's appropriate.
We earn our keep on services — deployment, customization, vertical configuration, and support — not by holding your data hostage.
Every module shares one data boundary and one audit trail — not ten policies to reconcile. Designed from day one to address GDPR, PDPO, PDPA and the EU AI Act, for SMBs in regulated industries.
- ⭐ Star the modules you find useful — it genuinely helps.
- 🐛 Open an issue for bugs, ideas, or questions.
- 🔧 Contribute — PRs welcome; check each repo's contributing guide.
- 💬 Need help deploying? Reach out at hello@odw.ai.
Own your AI. Start in the open.
© 2026 ODW·AI · OnDemandWorld