# Improve with feedback

Use task outcomes and feedback to guide the next choice of Plugin or API.

SpringBrand can review the result of a selected API or Plugin and refine its next choice. Tell your Agent what the result helped you decide, what was missing, and what should change on the next run.

## Review a completed task

- Did the result answer the original task for the right audience?
- Which source, asset, or action was useful, and what evidence was missing?
- Would the next run need another capability, different inputs, or a narrower question?

## Give useful feedback for the next run

Keep a short record of the goal, operation, observed result, and your correction. Provide it with a follow-up request when you want the Agent to use that context. Do not assume that a separate Agent session retains your previous task details.

## Try a follow-up request

**Copy a follow-up prompt**

```text
My previous SpringBrand task was [goal]. It used [operation] and returned [result]. I found [what helped] and [what was missing]. For the next run, prioritize [new constraint or question]. First inspect the available operations and explain which one fits. Do not run a paid or external action without my approval.
```

> This page describes using results and explicit feedback. It does not promise persistent memory across Agent sessions.

- [Smart tool matching](/developers/docs/tool-routing): See how tool selection and refinement work.
- [Keep working 24/7](/developers/docs/continuous-execution): Check how a task may continue beyond an Agent session.