XENN is built on open, self-hostable technology. Three layers carry the work: an agent runtime, a reasoning layer that improves itself, and a workflow engine that turns decisions into actions.
Open-source, local-first AI agent runtime. A single gateway process connects the chat channels you already use to a skill system and the model of your choice — conversations, memory and credentials stay as plain files on hardware you own. MIT licensed.
openclaw.ai ↗Nous Research's agent framework and the successor to OpenClaw: the same personal-assistant pattern, plus a closed-loop self-improvement layer and goal-oriented planning. XENN's own agents run on this layer.
nousresearch.com ↗Open-source workflow automation. APIs, databases and messaging are wired into deterministic, auditable pipelines — the hands that carry out what the agents decide.
n8n.io ↗Self-hosted by design. Prompts, memory and keys live on infrastructure you control — not in somebody else's account.
Model choice stays open. Agents and workflows are decoupled from the model provider, so the engine underneath can be swapped without rebuilding the system.
Automation you can audit. Agents decide, workflows execute — each step is deterministic and leaves a trace, so results can be checked, not just trusted.
Create a free account and call the models directly — or talk to us about wiring agents and workflows into your operation.
XENN