Skip to content

Node Failed to Learn

Whenever you save information to myReach, the AI automatically begins the "Learning" process. You can monitor this in the Properties tab of any Node. While most items process in seconds, you may occasionally see a status of "Not Learned" or "Failed to Learn".

Here is how to resolve these two distinct scenarios.

Not Learned

If a Node is marked as "Not Learned", it typically means it is currently in the processing queue. This is common if you have just performed a large bulk upload or synced a significant number of files from an integration.

What to do:

  • Wait: Give the system a few minutes to catch up. The speed depends on the complexity of the file (e.g. a long video transcript takes longer than a text note).

  • Manual Trigger: If the status does not change after several minutes, you can push it manually. Open the Node, navigate to the Properties tab, click on the AI Status and select "Learn" from the menu.

Failed to Learn

A "Failed to Learn" status indicates that the AI encountered a specific error while trying to process the file. This could be due to a temporary server interruption, an unsupported file format or a corrupted file.

What to do:

  • Retry: Click on the Node and go to the Properties tab. Click on the "Failed" status and select "Retry". In many cases, a simple retry resolves temporary connectivity issues.

  • Verify File Type: Ensure the file is one of our accepted formats. Supported types include PDF, DOCX, XLS, PPTX, TXT, images (JPG, PNG), audio (MP3, WAV) and video (MP4).

  • Check File Quality: If you are performing OCR, ensure the image is clear and high-resolution. If the AI cannot detect any text, the learning process may fail.

  • File Size: Verify that the file does not exceed the Storage Limits of your current plan.

Pro Tip

If a website fails to learn, it might be behind a "paywall" or require a login that the myReach AI cannot access. In these cases, we recommend using the Browser Extension to save the page content directly while you are logged in, or using "client-side rendering" in the integration setup page, to bypass these blockers.