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

# Save a handoff checkpoint

> Save the current task state and next action in a memory you can find again.

<div className="course-meta" aria-label="Lesson information">
  <span><Icon icon="book-open" /> AI Workflow · Lesson 3 / 5</span>
  <span><Icon icon="clock-3" /> \~5 min</span>
  <span><Icon icon="user-round" /> Intermediate</span>
</div>

Before leaving the first AI tool, save what the next session needs: the current draft, confirmed decisions, open questions, and next action. This concise memory is the handoff checkpoint.

## What you'll accomplish

<Check>
  You'll save a handoff checkpoint in Mem and find it again.
</Check>

## Watch how it's done

<Note>
  The demo uses Cursor as an example.
</Note>

<iframe src="https://www.youtube.com/embed/Z0mGaPfT6GQ?autoplay=0" title="Demo: save a handoff checkpoint" allowfullscreen="true" />

## Before you start

Keep Mem running and have the outline from [lesson 2](/ai-workflow/start-with-your-first-ai) ready. Use the same task name so you can find the brief and checkpoint together.

## Your turn

<Steps>
  <Step title="Review the first AI's output">
    List what you accepted, what remains open, and the exact next action. You can ask the first AI to draft the checkpoint:

    ```text theme={null}
    Draft a concise handoff checkpoint for the Atlas landing page. Include the current six-section outline, decisions I confirmed and their reasons, unresolved questions, and the next action. Keep suggestions I have not accepted marked as undecided.
    ```

    Read the draft before saving it. Correct anything that turns an unconfirmed suggestion into a settled choice.
  </Step>

  <Step title="Save the checkpoint as a memory">
    For this exercise, open **Timeline** in Mem. Adapt this example to match your actual output and decisions, then paste it into the input and press Enter:

    ```text theme={null}
    Atlas landing page handoff checkpoint

    Completed: reviewed the six-section outline below.
    1. Hero: less status chasing.
    2. Problem: scattered team updates.
    3. Workflow: weekly planning.
    4. Proof: an example supporting the workflow.
    5. Fit: who the product is for.
    6. Call to action: try Atlas.

    Decision: use "Try Atlas" as the single call to action throughout the page.
    Why: it gives readers one clear next action, as required by the work brief.
    Still open: choose a verified example for section four; no customer quote or metric is approved yet.
    Next step: draft the hero and section four copy. Mark missing evidence as "needs evidence" instead of inventing a claim.
    ```

    <Tip>
      Include the part of the draft needed for the next step. If the artifact is long, record where it lives and confirm the next AI can access it. "The outline is finished" is not enough.
    </Tip>
  </Step>

  <Step title="Ask Mem to find the checkpoint">
    Confirm the saved entry appears in **Timeline**. Then ask:

    ```text theme={null}
    What is the latest handoff checkpoint for the Atlas landing page? Include the outline, confirmed decision, unresolved proof point, and next action.
    ```

    Submit the question and compare the answer with what you saved. Check that the missing proof is still described as unresolved.
  </Step>
</Steps>

## What you should see

<Check>
  Mem brings back the checkpoint, including the outline, the agreed call to action, the missing proof, and the next writing task. You have enough context to resume without reopening the first chat.
</Check>

<Note>
  If the answer misses your checkpoint, confirm it was submitted, then search for a distinctive phrase such as "Atlas landing page handoff checkpoint". Open the saved memory and check its content before trying the question again.
</Note>

## Why this matters

The work brief records the goal and constraints. The checkpoint records where the latest pass ended. Together, they give the next AI the accepted decisions, open questions, and next step.

## Go deeper

Read [Memories](https://mem.nowledge.co/docs/memories) for other ways to create, find, and edit reusable context.

## Next lesson

<p className="course-status-line"><i aria-hidden="true" />Coming soon</p>