Skip to content
Nappai Documentation

In-Memory Checkpointer

The In-Memory Checkpointer acts as a temporary storage box for your workflow’s progress. It saves where you are in the process so you can pick up right where you left off, but only while the automation is running. Once the automation finishes or restarts, the saved state is cleared. It is like a ‘draft’ saved in your brain while you work, rather than writing it on paper that stays forever.

How it Works

When you run this component, it takes the current data and extra details and stores them in the system’s fast memory. This saves your place in the workflow. Later, you can retrieve this saved data using a unique name. This helps your automation pick up exactly where it stopped, making the process faster and allowing you to manage interruptions easily. The component is designed for short tasks and debugging, ensuring quick access to your data without slowing things down.

Inputs

The following fields are available to configure this component:

  • State: The current status or data of your workflow that you want to save.
  • Metadata: Extra information like timestamps or IDs that help track and organize the saved checkpoint.
  • Checkpoint Key: A unique name or ID to identify this saved checkpoint so you can find it later.

Outputs

The component produces the following results:

  • Saved Checkpoint: The data you saved, ready to be passed to other steps in your workflow or used to resume a process.
  • Operation Status: A message indicating whether the save or load operation was successful.
  • Updated Metadata: The metadata after the operation, useful for tracking versions and debugging.

Output Data Example (JSON)json

{ “Saved Checkpoint”: { “context”: “Current workflow state”, “data_snapshot”: ”…” }, “Operation Status”: “success”, “Updated Metadata”: { “timestamp”: “2024-05-20T14:30:00Z”, “version”: “v1” } }

Connectivity

This component is typically connected to other steps in your workflow where you need to remember the progress. For example, it works well with AI agents or tasks that run in loops, allowing the workflow to resume from a specific point. Connect the State input from previous nodes to define what to save. Connect the Saved Checkpoint output to downstream nodes that require the stored context to continue the process.

Usage Example

You have a workflow that processes a list of emails. You use the In-Memory Checkpointer to save your progress after every 10 emails. If the workflow stops unexpectedly, you can retrieve the saved checkpoint and continue from where you left off, ensuring no emails are missed. The Checkpoint Key ensures you save different batches separately.

Tips and Best Practices

  • Temporary Storage: Data is cleared when the session ends. Do not rely on this for permanent records.
  • Unique Keys: Use unique Checkpoint Key values for different tasks to avoid conflicts.
  • Performance: Ideal for fast iterations and debugging due to memory-based storage.
  • Development Status: This component is currently in development. Features may evolve, so check for updates.

Security Considerations

Since data is stored in memory only, it is not written to disk, which can reduce persistence risks. However, because this component is in development, review sensitive data handling as features stabilize.