This article explains when Tensor Admin asks you first, and how to let routine work run without asking.
The default: Tensor asks first
Tensor asks before it writes or sends anything. A change to your records, or a message going out, waits for you.
A reply or message Tensor writes always waits for approval.
What you can do with an approval
When a run is waiting, you have three choices:
- Approve. The step goes ahead as proposed.
- Decline. The step does not happen. A declined run costs nothing.
- Correct. Change what Tensor proposed. The correction updates the automation's instructions as a new version, so the next run does it your way.
Limits
Once an automation gives steady results, approving every routine step becomes slow. Limits solve that.
You set limits Tensor may work within. Routine steps inside those limits run without asking. Anything outside them waits for you, as before.
A sensible way to set limits:
- Run the automation with full approvals until the results are steady.
- Notice which steps you approve every time without changes.
- Set limits that cover only those steps.
- Read receipts for the next few runs to confirm that the right steps ran on their own.
- Widen the limits later if you want to. Narrow them if something surprises you.
When Tensor stops
When information is missing or conflicts, Tensor stops instead of guessing. This happens whether or not limits are set.
- Open the run and read what it could not work out.
- Supply the missing detail or settle the conflict.
- If the same gap keeps coming up, correct the automation so it knows what to do.
Seeing what was approved
Every run leaves a receipt. It shows what came in, each step Tensor took, what you approved, and the result. Use it to check which steps ran inside your limits and which waited.
Undoing a change
Each automation keeps versions. If a correction makes results worse, roll back to an earlier version.
Who approves
Team roles and permissions exist on every plan.
Related
- Troubleshooting covers runs that are waiting or stopped.
- Security and privacy covers how sign-in details and workspace data are protected.