Added knowledge source and testing if the Agent could read the PDF files in a Dataverse table row but it just gets the file guid, not the contents so knowledge source is not a choice now. I need to create Agent Flow that gets the PDF file from Dataverse row and then use AI Builder to read the CV contents for the Agent to use it for comparison. The CV file uploading instructions are in previous post.
First I tried to add skill in the end of the job applying topic but get stopped because missing license. It seems skill is not what I need here, since I don’t have another agent doing the work for me.

Then I tried to create a Tool in the Agent. I set the properties like in the below picture.


I just don’t know how to use it
I did not get the tool working. I tried to change the properties but eventually figured out that Tools is not the correct technology for getting file contents – or I just don’t know how to use it.

Then I decided to create Agent Flow that extracts the PDF file content but noticed that I need to training AI Builder model first for the Process Document action. I could do this also when saving the document into Dataverse if I create new column CV content or something.

When going trough models in AI Builder I noticed that there were already ready made model for extracting text from picture or PDF. I used the recognition action and then set the full text into the variable which you pass back to the agent in the last action.

And finally I succeed to get the CV content from the file to the Agent.

Then it was time to update the instructions for the Agent

Now my Agent can use the CV content from a file in Dataverse custom table. But the agent does not understand that it got the CV content already. Handling this issue is described in the next post to come.

In next post I will compare the file contents with the job position description.