Schedule an AI Agent

Video walkthrough

What this does

A scheduled agent is an existing AI agent given a specific task to run on a recurring schedule - daily reports, threshold alerts, custom SQL runs, and so on. Each scheduled run uses the agent's connection set, integrations, and CLI access, and posts results to its messaging integrations or the Bruin Cloud chat.

If you haven't created an agent yet, see Configure AI Agents first.

Steps

1) Open scheduled agents

From the AI menu, navigate to Scheduled agents.

2) Create a new scheduled agent

Click New scheduled agent, give it a name, and pick the underlying agent it should use. The schedule inherits whatever connections, integrations, and CLI access that agent already has.

3) Configure the task

The configuration page has two halves:

• Right panel - the actual settings: name, schedule, instructions, notifications, custom SQL, output format.
• Left panel - a chat where you can describe the task in plain English and let an AI build the configuration for you.

Either set things up manually or describe the task and let the agent fill it in.

4) Set the schedule

Pick from built-in presets, or write a custom cron expression for full control.

5) Set up notifications

Pick a notification channel - Slack, Teams, WhatsApp, or any other integration that's already wired up to the underlying agent. Only integrations the agent has access to will appear here.

6) Optional - custom SQL or output format

You can manually:

• Provide a SQL query the agent should run.
• Specify an output format - for example a PDF report, a formatted Slack message, or a threshold-based alert.

Example - daily stock report

In the video, the prompt to the configuration chat was:

"Send a daily PDF report to our Slack channel summarizing the previous day's Apple and Microsoft stock. If a stock moved more than 5%, format the message as an alert."

After about a minute and 10 steps, the agent built the full configuration:

• Wrote the instructions for the scheduled run.
• Set the schedule to a daily cadence.
• Set the notification to the Slack channel.
• Generated the SQL query to run each day.
• Set the output format to a PDF report plus a Slack alert when the threshold is hit.

Activate and run

Once the configuration looks right:

• Enable the scheduled agent. You'll see the next run time (UTC).
• Use New run to trigger it manually any time.
• Open the Runs tab to see history and status, and jump into any specific run.

Where the runs live

Scheduled agents execute inside the AI Chats view. That means:

• A scheduled agent with no messaging integration only shows up under Chats.
• A scheduled agent with an integration (Slack, WhatsApp, Teams, etc.) still appears under Chats - the integration is just the destination for the output.

Key takeaways

• Scheduled agents reuse an existing agent's connections, integrations, and CLI access.
• Build the schedule manually, or describe the task in plain English and let the AI configure it.
• Cron expressions are supported for custom schedules.
• Output can be PDFs, formatted messages, threshold alerts, or any combination.
• All runs - manual and scheduled - show up under the Chats view, regardless of integration.

Next

To wire up the messaging integrations a scheduled agent can post to, see Configure AI Agents. To use the agent ad-hoc instead of on a schedule, see Chat with an AI Agent.