What Autopilot does not do

It does not share outside the project

A workflow belongs to one project. There is no organization-wide view, and you cannot share a workflow to another project. To use the same workflow in two projects, download it and import it into the other one.

Nothing in Autopilot is public. Forms and chats you build are open only to signed-in members of the project. Publish means live in this project, not visible to the world.

It does not go live without you

Atlas can draft and edit workflows, but it never publishes one or turns it on without your approval. A draft never runs on its own. Only a published version fires its triggers.

It does not decide what is good

A workflow does exactly what its nodes say. It will not judge a creative unless you add a Quality check node. It will not pause a weak ad unless you build that rule with Get performance, If, and Pause ad. If a step needs taste, put a person in the loop with Send and wait for response.

It does not run tests for free

Test runs are real. AI steps in a test use credits, and posting or sending steps really post and send. Pin data or deactivate nodes while you build.

It does not keep data forever

Execution data is removed after 30 days at most. If you need a record, send it somewhere you keep, such as a data table, a Slack channel, or an email digest.

It does not run without limits

Each live workflow has fair-use limits: 2,000 executions, 30 minutes of Code node time, and 5,000 HTTP requests a day. Schedules can fire at most every 5 minutes. A workflow that fails three times in a row is unpublished. A workflow that cannot pay its daily fee is unpublished too.

It does not install outside packages on Pinnora servers

n8n lets you install community nodes onto your own server. Pinnora is shared, so it does not. Instead, you can save a Code node with a settings form as a custom node for your project. It runs in an isolated sandbox, not on Pinnora's servers.

What it does do well

Within those limits, Autopilot runs your routine work: posting on a schedule, making a weekly batch of ads, sending a performance digest, alerting your team, and pausing for approval where it matters. Build it once, test it, publish it, and check the Executions tab when you want to see how it went.