Skip to main content
Rate the answers you disagree with, right from the cell that holds them. Then run a training round: Oneprofile rewrites the prompt to fix what needs work, and you keep the change before it ships.
The Train screen mid-round: a Round 1 chip beside the agent's name, a progress strip counting rows marked, and one answer's card with its Input and Steps taken rows above Good and Bad buttons.

Reviewing one example during a training round.

​
Rate a cell any time

Right-click a cell, or a selection of them, and choose Looks good or Needs work. Or open the cell and use the same two buttons at the bottom of its drawer. A note on a Needs work rating is optional there, but it is the detail a rewrite can act on. In the drawer, click Add note to say what was wrong.

​
Run a training round

1

Open Train

Open the agent and click the Train tab, or Train agent in its Training card. The row menu on the agents list has Train agent too.
2

Choose training data

Pick Enter manually to type example rows, or Use a table to run against real ones. Click Run examples once your rows are ready.
3

Check each example

Click Looks good, or click Needs work and say what is wrong and what it should have returned instead. Once every example has an answer, the summary counts them.
4

Decide on the change

Click Teach the agent, then read What will change. Click Keep this change, Try a different note for another draft, or Discard.
Needs work always needs a note. Looks good needs nothing and moves you straight to the next example. Press G or N instead of clicking. With at least one example marked Needs work, Teach the agent has Oneprofile propose a change to the prompt. Prompt changes under it shows the exact wording.
Keeping a change saves a new version and re-runs your examples. Every table column using this agent picks up the change on its next run.

​
Good to know

The examples and feedback are kept as history, but the round ends and the agent is free for a new one. A revision you already approved stays applied.
No. Discard drops only this round’s change and its notes. Feedback from earlier rounds is kept.
Mark it Needs work again with a fresh note. An example left unchanged since your last note is flagged, so you know it still needs attention.
Yes. Open Change effort from the training screen’s menu, without leaving the session. The next round uses whatever you save.

​
Next steps

Agent versions

See what an approved revision does to earlier versions.

Read an agent run

Look at the trail behind an answer before you decide it is wrong.