Your platform is ready
Check your deployment, choose what to automate next, and get help from Agno.
You now have a place to build agents, run them for your team, and improve them over time. Before handing it over, use this checklist to confirm that the parts you enabled work together.
Check your platform
Complete these checks on your deployed platform:
- Your agent does useful work. Run Daily Brief, read the edition, and check its dated original sources. Ask a follow-up in the same conversation.
- Your deployment is secured. Confirm
RUNTIME_ENV=prd, JWT authorization, and the public HTTPS URL using the deployment guide. Verify that an unauthenticated agent run is rejected. - Your daily schedule works. Check the executor, instructions, and timezone in Scheduler. Trigger a run and inspect the completed result. Keep one matching schedule.
- You can investigate a problem. Find a run's session, tool results, and trace in AgentOS, and find the corresponding service logs in Render.
- You can ship an improvement. Reproduce one weakness, compare a focused change, and retain a useful regression case using Improve your agents. Review the change before deploying it.
Check the capabilities you enabled
You can start with one agent on one API instance. Complete the following checks as you add these capabilities:
| Capability | Evidence that it works |
|---|---|
| Coding agents and sandboxes | A coding session uses the intended repository and branch, completes a small task, and opens a reviewable PR. Check checkpoint recovery too. |
| Scheduled improvements | A hosted run uses isolated test data, records before-and-after results, and produces a useful draft PR when a change is justified. |
| Company OS | A proposal cites readable Sources. You accept a Task, start its run, inspect its result in Reviewing, and mark satisfactory work Done. |
| Slack | A DM or channel mention reaches the intended agent and returns a reply. If you added scheduled delivery, verify the channel receives one edition. |
| Multiple replicas | Streaming, cancellation, scheduled work, and recovery work across instances with shared configuration. |
A configured feature is ready once you have seen its result. Leave unfinished automation paused while you complete its prerequisites.
Choose what runs automatically
Review Scheduler and the Routines on your Company Products. Confirm each enabled item's purpose, owner, executor, frequency, and timezone. Enable the maintenance Routine from the Company page in the AgentOS UI when you are ready to review its proposals.
If both a weekly improvement schedule and a Company Routine cover Daily Brief, choose which one owns that work. Keep automatic acceptance off while you learn what the Routine proposes.
To stop future work, disable the schedule in Scheduler or pause the Routine on its Company Product. Disabling future runs does not cancel a run already in progress; inspect and stop that run separately.
Where to next
Pick one recurring responsibility your team already spends time on. Give it a clear outcome, accessible evidence, and a result someone will review. Run it manually before adding a schedule.
- Make Daily Brief part of your team's morning. Refine its coverage with reader feedback and add Slack delivery.
- Keep your platform improving. Use Company OS to review changes and propose small, evidence-backed improvements.
- Build the next agent. Explore the examples, then use the
create-agentskill in your repository to build around your own workflow.
Add one responsibility at a time. Keep the same loop: inspect the result, give feedback, test the change, and decide what to adopt.
Get help from Agno
For implementation questions, ask in GitHub Discussions or join the Agno Discord.
For help planning your rollout, enterprise guidance, or professional support, email support@agno.com or book a call with the Agno team.
When reporting a problem, include the template and Agno version, what you expected, what happened, and the relevant sanitized logs or error. Include a small reproduction when possible. Remove API keys, tokens, and private data before sharing.
We would also love to see what you build. Share your workflow and what you learned with the community. See Getting Help for all support and community links.