Quickstart
Start using the AI Assistant in a couple of minutes. This short guide takes you from opening the assistant to running your first real action.
Prerequisites
- A Cordatus account with access to the Web App or Client
- The permissions required for whatever you want to do (the assistant inherits your account's roles)
- For live actions on a device: at least one online device and, for camera actions, at least one camera added
See Device Installation and Connecting Cameras if you have not set these up yet.
Getting started
1. Open the assistant
- Open the AI Assistant panel in the Cordatus dashboard.
- You will see a chat input with a prompt like "Ask Cordatus to do something…".
2. Type your first request
Describe what you want in plain language. For example:
"I want to watch a camera."
You do not need any special syntax — just say it the way you would say it out loud.
3. Answer with the cards
If your request could match more than one thing, the assistant shows selection cards. For the example above, it lists your cameras so you can pick one:
"Open the entrance camera."
Pick the camera (or refine by name) and the assistant opens the live stream.

4. Confirm sensitive actions
For actions that change state or cannot be undone, the assistant shows a confirmation (or a summary card) before it proceeds. Review it, then approve or cancel.
The assistant never performs a destructive or irreversible action without confirmation.
5. See the result
Once confirmed, the assistant performs the real action and shows the result — a live stream, a started recording, a running job, or a summary card.
Try these first
| You say | What happens |
|---|---|
| "Open the entrance camera." | Opens the live stream for that camera |
| "Record the entrance camera." | Starts a recording on that camera |
| "Start people detection on the entrance camera." | Launches an inference job with the right model |
| "Show the factory device status." | Displays live device metrics (CPU / GPU / memory) |
| "Show today's alarms." | Lists triggered alarm events for today |
See the full Example Prompts library for more.
Expected result
- The assistant understood your request and routed it to the right domain.
- Any ambiguity was resolved with a selection card.
- Sensitive actions asked for confirmation.
- The real action was performed and its result shown.
Next steps
- Learn everything the assistant can do in Capabilities.
- Browse ready-to-use commands in Example Prompts.
- Understand the design principles in How It Works.