> ## Documentation Index
> Fetch the complete documentation index at: https://docs.screenpipe.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Recording, memory, workflows, and automation: what is the difference?

> Screenpipe captures available screen and audio context while recording is running.

Screenpipe captures available screen and audio context while recording is running. An AI can retrieve that history, turn it into a report, or use it in an automation. Those are separate stages: a recording does not by itself prove that a report was saved or an action completed.

## Start here

Choose one short work session and one output, such as a reviewed job log. Search for a distinctive phrase from that session before asking for a summary.

## Copy this prompt

Use this in Screenpipe chat or an assistant with the required connections. Replace the brackets with your details. If the assistant lacks access, provide a reviewed excerpt; a prompt alone does not connect it to your history.

```text theme={null}
Use my recorded work on [date], [start–end, timezone]. First confirm what
history is available. Then explain the session using four sections:
1. Observed activity, with source times.
2. A concise memory worth keeping.
3. A possible repeated workflow, marked as a hypothesis.
4. One suggested automation and how I would verify its result.
Do not execute the suggestion. If the sources are missing, say what is missing.
```

## Check the result

Open a source reference. Check that a proposed workflow is labeled as a proposal. If an agent says it saved a report, open that file or destination and inspect its contents.

<AccordionGroup>
  <Accordion title="See a worked example" id="worked-example">
    | Stage      | Fictional example                       | Proof to look for                          |
    | ---------- | --------------------------------------- | ------------------------------------------ |
    | Recording  | A request was visible at 10:15.         | Search result or source moment.            |
    | Memory     | The team chose a manual approval step.  | Reviewed note with a source.               |
    | Workflow   | Approved requests become checklists.    | Operator-confirmed steps and exceptions.   |
    | Automation | Draft a checklist for the next request. | Actual saved output and a successful test. |
  </Accordion>

  <Accordion title="Go deeper: setup choices and edge cases" id="details">
    A **job** can mean your client assignment, while a **scheduled task** is a configured AI task. Give client jobs stable labels in your notes and prompts. Do not assume that naming a job creates a built-in billing record.

    A voice request is another way of expressing intent. It only controls recording, a running AI task, or a schedule if the assistant has the corresponding tool and successfully uses it. Ask it to identify the target and report the resulting state. Stopping a run and disabling future runs are different operations.
  </Accordion>
</AccordionGroup>

## Continue

[First useful result](/first-useful-result) · [AI usage and controls](/ai-usage-and-controls) · [workflow handoff](/workflow-handoff)