Assistance Needed – Copilot Studio Knowledge Files Not Processing

Question

Hi,

I'm testing Microsoft Copilot Studio to build an AI assistant for the Dexter Southfield Athletics Department.

The agent itself was created successfully, and public websites are added as knowledge sources without any issues.

However, whenever I upload a document (Excel, Word, or PDF), the file either remains "In progress" indefinitely or eventually changes to "Error." This happens even with very small test files, so it doesn't appear to be specific to one document or file type.

Could you please check whether there is anything in our Microsoft 365 environment that would prevent uploaded files from being indexed in Copilot Studio? Specifically:

• Is my Copilot Studio/Microsoft 365 licensing sufficient for document knowledge sources?
• Is the Dataverse/Copilot Studio environment fully provisioned?
• Are there any tenant, security, or compliance policies blocking uploaded files from being processed?

If everything appears to be configured correctly, would you be able to open a support case with Microsoft to investigate?

I'm happy to provide screenshots or additional details if needed.

Thank you!

Answer 1

The supported checks from Copilot Studio documentation point to these likely causes for uploaded document knowledge sources staying In progress or moving to Error:

1. Verify the knowledge source status meaning in Test your agent.
   • In progress means the file upload knowledge source isn’t ready yet.
   • Error can be caused by:
     • unsupported file type or size
     • knowledge source limit reached
     • encrypted file or sensitivity label
     • product or system error
   • For a product or system error, the documented action is to delete the file and reupload it.
2. Check file restrictions that block processing.
   • Files with Confidential or Highly Confidential sensitivity labels aren’t supported.
   • Password-protected or encrypted files aren’t supported.
   • For uploaded user files, encrypted or protected Office files can return an explicit encrypted/password-protected message or a generic error.
3. Confirm the file type is supported for the scenario.
   • For SharePoint upload knowledge sources, supported types are: doc, docx, xls, xlsx, ppt, pptx, pdf.
   • If using file input to chat rather than knowledge-source upload, supported types differ and include DOCX, CSV, PDF, TXT; XLSX and PPTX are only available in experimental mode there.
4. Check knowledge object limits.
   • An agent can’t have more than 500 knowledge objects. If that limit is reached, more objects can’t be added until previous ones are deleted.
5. If the files are from SharePoint or OneDrive, confirm access and readiness.
   • A file can appear in the knowledge source but still not answer if the source isn’t Ready, the file has unsupported characters in the name, the file type isn’t supported, the file is protected, or the file wasn’t shared appropriately.
6. If the source is SharePoint, note the documented ingestion behavior.
   • After adding files or folders, status can show Ready immediately, then change to In Progress, and only when it returns to Ready is the content ready for use.
   • SharePoint synchronization frequency is four to six hours based on ingestion completion.
7. Check environment conditions that can block knowledge usage.
   • Unknown status can occur when Dataverse search isn’t enabled, authentication is incorrect for Dataverse table knowledge sources, or a data policy blocks the knowledge source.

The context does not provide support for checking tenant licensing sufficiency, confirming whether the environment is fully provisioned, or opening a Microsoft support case.

---

• Test your agent's knowledge sources
• Copilot Studio quotas and limits
• Unstructured data as a knowledge source
• Allow file input from users
• Allow file input from users
• Test your agent's knowledge sources