The model is headless. This is what one UI on it looks like.
The Large Behavior Model runs behind a REST API and MCP tools, so agents consume it without any interface. This live preview is the Platform MVP: a working dashboard built on those same APIs. The first use case it ships is agent routing; every panel in it is just a different probe of the same model.
Two ways in
Both run on anonymized data: people appear as codes like ENG-01, never names. There are also role-based product walkthroughs for AI teams, ops, change leaders, and governance.
Guided demo
A scripted walkthrough with three storylines: a platform team unblocking stalled agents, an ops leader reading live org health, and a governance team auditing AI decisions. Auto-plays, six scenes each.
Start the demo →
Live network map
The interactive organizational graph behind the dashboard. Switch between relevance, routing, and risk views, hover any person to inspect load, capacity, and bridge roles.
Open the map →
One model underneath. The UI is just a probe.
The Platform MVP's first app is agent routing: it shows who work should go to, who is overloaded, and where escalations land. But the panels around it (bottlenecks, guardrails, escalations, adoption, change network) are reads of the same representation. Point the UI at a different probe and the same dashboard does knowledge governance, workforce risk, or change management.
- Agent routing: the flagship probe, shipped first
- Bottlenecks and team health: the ops view of the same graph
- Guardrails and audit: the governance view of the same decisions
- Adoption and change network: who actually drives a rollout
// Everything the UI shows comes from the same API POST /v1/agents/route ← routing panel GET /v1/bottlenecks ← ops panel GET /v1/team/{id}/health ← team view POST /v1/chat/ask ← chat thread // Or skip the UI entirely: your agents // call the same endpoints over MCP behaviorgraph/route behaviorgraph/team_health
Want this running on your organization?
Connect permissioned data, evaluate the model on your own org, then consume any capability through the REST API, MCP tools, or a UI like this one.
Get a demo Read the routing use case