Introduction
henosis gives your operation one shared, live picture of every asset — one that your people, your apps, and your AI agents can all use to get answers in seconds and act on them. Your data never moves; henosis connects the systems you already run and makes them coherently queryable by humans and AI alike.
It is open-source and self-hosted under MIT.
One opinionated model
henosis's ontology isn't a blank graph of entities and relationships. Whatever you're reasoning about — a well or a pump, but equally a maintenance schedule, a work order, or an event — a decision about it draws on the same few things:
- Identity — what it is
- Location — where it is
- Observation — what it's doing, now and over time
- Documents — what we know about it
- Relationships — what it's connected to
You combine them into a decision and close it with an action that changes the asset, feeding the next one. Because henosis builds these in as first-class roles — not just generic nodes and edges — it knows an observation is a time-series and a location is a place. So it can drive a map, a trend chart, or an agent's tools straight from your model. A generic ontology can't: it's a blank canvas you have to build all the meaning on top of.
Three surfaces, one model
The same compiled model is served three ways, so a person, an application, and an AI agent all reason over exactly the same context:
- The Canvas (coming soon) — a visual workspace for people: maps, trends, entity detail, and agent-arrangeable panels.
- A REST API (
/v1) — for applications and systems. - MCP tools (
/mcp) — for AI agents, exposing the model as typed tools they can call.
What it's not
henosis sits on top of your data foundation — it doesn't replace the work beneath it. It is not a data warehouse, a semantic/metrics layer for BI, a dashboard, or a generic data catalogue. It is a live, queryable, opinionated model of your assets that people and agents can act on. Your systems of record stay authoritative; henosis stores schema and routing only, never your data.
Where to go next
- Quickstart — install henosis and serve the worked example in a few minutes.
- Architecture — how the three components fit: the ontology you author in YAML, the engine, and the Canvas.
- Ontology & data sources — author your own model in YAML and point it at the systems you already run.