View API-Uploaded Documents in the Mindee Interface

We are very interested in Mindee and are currently evaluating your solution.

At the moment, we are using Rossum in production, and we are testing Mindee to compare both platforms and potentially switch if the results are successful.

There is one important feature that is currently missing for our workflow:
the ability to view documents uploaded via API directly in the Mindee interface, similar to how documents can be reviewed in the Live Test.

This capability is very important for our operational and support teams, as it allows easier verification, troubleshooting, and collaboration when processing documents programmatically.

We would really appreciate it if this feature could be considered for your roadmap.

Thank you for your work and support.

Best regards,

Posted by vtp · Mar 24

Completed3

Ianaré Sévi

Team•

Mar 24

Hi, thanks much for the feedback.

This is possible, it would obviously mean storing the documents and their results on the Mindee servers. Most of our users do not want this, so if implemented it would be an option to pass in the API call.

I’m curious as to how you would use the documents though… what kind of interface and data would be needed for your use case? Is the current “history” view enough or do you need some other features?

Thanks!

vtp
•
Mar 24
Thank you for your response.
We need the ability to review documents and correct the extracted data results, similar to the RAG workflow, and we also need a clarification regarding the RAG feature.
We are currently testing Mindee with a trial account, so we do not have access to the RAG functionality.
If we subscribe to this option and activate RAG via API calls, will we be able to find, review, the documents posted by api in the “Continuous Learning (RAG)” section of your interface?
Thank you in advance for your clarification.
- Ianaré Sévi
  Team•
  Mar 25
  The RAG system has 2 parts: the database and the inference/processing.
  The database (“RAG documents” on the platform) is where documents are stored, and where you can validate and adjust as needed. These adjustments are only within the context of the RAG system. These documents are not sent via API, they need to be manually uploaded on the platform.
  For the inference/processing, the files are sent via API, and the documents in the RAG database are used to enhance/correct the files sent by API. These files are never stored on the servers.
  Basically the usual process is that nothing is stored on Mindee servers except for the documents used to “train” the RAG system.
  But again we could implement a feature where you can set an option to store a file when making the API call.
  Just to get an idea of the needed storage on our end, how many files do you plan on sending and storing per week?
  And also, is your intent to have a manual validation step for each file sent? Basically a process where you send files, they are reviewed manually on the Mindee platform, any corrections are made directly inn Mindee, and finally sent to your system?
  In any case, I’ll bring up this in our next feature planning meeting.

View API-Uploaded Documents in the Mindee Interface

Comments3