Follow the steps
Create a blank automation
Open Runs → New automation → Blank automation, then choose Create automation. Give it a name that identifies the workspace and the daily review.
Runs → New automation (opens in a new tab)Set the morning in your time zone
Select the trigger and set Trigger type to On a schedule, Cadence to Daily, and the desired Hour. Choose your Time zone explicitly so the scheduled hour has a clear meaning.
Run builder → Schedule (opens in a new tab)Describe the review and its destination
Add an AI task with tools and choose Write my own instructions. Name the workspace, the previous calendar day in your chosen time zone, and the counts you need. Ask for call metrics for counts, rather than counting a paginated transcript list. Name the exact existing Chat conversation and explicitly authorize posting the report there.
Run builder → Add step → AI task with tools (opens in a new tab)Test the Run before enabling it
Review Available capabilities and save the Run. Choose Run now, complete any required confirmation, and inspect Run history. Check the reporting dates, totals, and destination before choosing Enable automation.
Runs → Run now → Run history (opens in a new tab)Check the actual team message
Open the chosen conversation in Scribed Chat. Confirm the report arrived once, names the reporting window and time zone, and contains the information your team expects. Return to Runs to inspect later executions.
Scribed Chat → your team conversation (opens in a new tab)
Check the result
- Run history shows the test’s steps and output.
- The selected Chat conversation contains the report with the expected dates and totals.
- The saved daily trigger shows the intended hour and time zone, and the automation is enabled.
Before you finish
- This is a Run you configure, rather than a prebuilt daily-review template.
- The notification step sends an in-app bell notification. Use the AI step’s Chat capability for a conversation post.
- Run permissions, confirmations, and usage admission still apply. Review a failed execution before relying on the next report.