AI Copilot
The AI Copilot is a conversational and agentic AI layer built into Cloud Studio's Gear platform. Operators ask questions in natural language and the Copilot answers using your real telemetry, generates dashboards on the fly, and executes actions on devices when authorized.
What you can do with the Copilot
- Ask about your data: "Show me the average temperature of the cold-storage fleet last week" → tabular answer + auto-built chart.
- Build dashboards by description: "Create a dashboard with energy consumption per facility and a SCADA view of pump #3" → pre-filled draft you can save.
- Run agentic actions (with permission): "Restart the gateway in plant Madrid-02" or "Acknowledge all critical alerts older than 24 hours."
- Investigate anomalies: "Why did the alarm fire at 03:14 on device DLR-117?" → root-cause summary citing the relevant endpoints.
- Generate scripts: "Write a low-code script that normalizes payloads from vendor X" → ready-to-paste snippet for the low-code scripting tools.
Where it lives
The Copilot appears as a sidebar panel inside Monitor. It is also available as a slash command inside the dashboards editor and the SCADA views.
Capabilities
| Capability | Status |
|---|---|
| Conversational query over telemetry (endpoints, devices, facilities) | GA |
| Auto-generated dashboards and charts from a prompt | Beta |
| Agentic actions on devices (command dispatch, alert ack, automations) | Beta — requires copilot.execute permission |
| Natural-language alert investigation | Beta |
| Multi-tenant scoping (results are filtered by your client permissions) | GA |
| Voice input | Roadmap |
Example prompts
Top 5 facilities by water consumption this month.
Alert me when any tank under category "Diesel" drops below 15% in the next 24h.
Why is gateway GW-204 offline since yesterday?
Build a dashboard that compares energy KPIs across the East region.
Run a low-code parser stub for protocol Sigfox 0x0A.Permissions and data scope
The Copilot only sees data the requesting user is already allowed to read. Agentic actions require the explicit copilot.execute permission on the affected client. All Copilot interactions are logged in the audit trail described in maintenance.
Limitations during beta
- Aggregations over time ranges longer than 90 days are throttled.
- Agentic actions are limited to a curated allow-list of commands per device type.
- Voice input and image attachments are not yet available.
- Generated dashboards open as drafts — review before saving.
Roadmap
Voice mode, multi-step workflows ("agents"), and proactive notifications are on the roadmap. Refer to the release notes for the latest GA status.
For a broader take on how AI and IoT combine into AIoT, see our AIoT pillar in the hub.
HttpResponse
The HttpResponse object allows returning data when sending uplink data through HTTP. Properties statusCode (int) The statusCode property allows indicating...
Requirements and Best Practices
This section applies only to cases where the platform needs to be installed on third-party servers (On-Premises). Minimum infrastructure requirements...