PDF as variable/text to be submitted to LLM

MindStudio IDE Feature Requests
1
  1. What problem does this feature request solve?

Ideally there should be a way to treat uploaded PDF as a variable, since the impression given it is possible. As I understood behind the scenes PDF uploaded to MindStudio cloud and represented as a link.

image
image1858Ă—1456 186 KB

image
image1886Ă—1376 338 KB

When you run it, it admits that it cannot read external links

image
image2116Ă—798 128 KB

Based on OpenAI api you can submit files similarly to GUI chat. I do understand it might be not the best from privacy perspective, so maybe a choice can be given to a user.

  1. What is the use case for this feature?

Use Case:

Agent to update portion for resume based on the 2-3 main features hiring managers are looking for based on the job description. Input - PDF resume, output either text or PDF

  1. Please describe the functionality of this feature request.

The functionality of this feature would be similar to uploading files to all LLMs chats, when you type in request and as a supplementary data upload the file.

E.g. screenshot from ChatGPT

image
image1498Ă—562 23.9 KB

  1. Is there anything else we should know?

Feel free to cancel this request if there are some privacy, security concerns or if there is some alternative way to achieve the desired result.

2

Hi @anastasiialysikova,

Thanks for the feature request.

Just to clarify, when you upload a file through a User Input block set to Upload File, all text is automatically extracted and saved to the defined variable unless the Processing option is set to No Processing.

Could you share a link to the Run from your third screenshot? I’d like to take a closer look.

3

@Alex_MindStudio here is agent ID, if this is what you mean 5b7fdf79-8b29-48ee-b42f-d3fe2a7572c0

4

Hi @anastasiialysikova,

Thank you for sharing the Agent ID!

It looks like the test value in the User Input block was set up incorrectly, so the file couldn’t be processed.

Can you try clearing the Test Value field in the Agent and make sure the Processing setting is set to “Extract Text”?

image
image757Ă—713 29.7 KB

Let me know how it goes!

5

Thanks @Alex_MindStudio , worked.

1 Like