AI Agents
Agents surface controller intelligence monitors for energy, comfort, and maintenance (plus related signals such as occupancy and levels). Use this page to watch scores, review recommendations, and set how aggressively agents may act.
:::tip Agents vs Assistant
Agents = dashboards, evals, interventions, and mode settings.
Assistant = conversational text chat that can answer questions and propose actions.
See AI Assistant.
:::
Open Agents
- Go to Agents in the sidebar.
- Choose scope: Site → optional Building → optional Floor / Zone.
Floors and zones come from the Facility snapshot of controllers linked to the selected building.
Tabs
| Tab | Purpose |
|---|---|
| Status | Monitor scores and trends (energy, comfort, maintenance, and related gauges) |
| Evals | Per-agent evaluation / rollup handling across controllers |
| Actions | Recommended interventions (generate, review; execution to the edge may be limited depending on release) |
| Settings | Operating mode and related policy (for example comfort minimum) when your role can edit |
Operating modes
Agents can run with different levels of autonomy (exact labels depend on server policy), typically along the lines of:
- Monitor only — observe and report
- Recommend — suggest interventions without applying them automatically
- Active — more autonomous optimization where entitled and approved
Pro (and Trial-as-Pro) licenses unlock AI agent capabilities; Basic focuses on core control without AI orchestration. See Licensing.
How scope rolls up
- Building membership includes all controllers assigned to that building.
- Site and building scores are typically the mean of membership controller scores for that monitor kind.
- Controllers with no building stay on the site inventory but are excluded from building-scoped rollups until assigned.