AI Center - ML Skills


I am trying to train a ML Model to process Purchase Orders.

I’m trying to get the MLSkill set up in studio, but when I select Robot there are no MLSkill available to select.


Also when I select Endpoint, I put in the URL after making the MLSkill public, but in all of the examples that I’ve seen there is normally an API key but there is no API key available on the ML Skills page.

Should I use the API key from the license?

Thanks for the help

Hi @samuel_trahan ,

Yes you would have to use the API key from the License.



  1. First check if the robot is connected to right orchestrator
  2. For endpoint try using the api from license page


Thank you

That’s what I was assuming.

I have another question


The input type is currently have is JSON but the file that I am inputting is a pdf. Should I change the filetype to File instead and then it was return a json response?


Yes please dependin on what ml needs you can select input as file or json…as it is pdf use file and send path

The output is always json


Thanks for the quick reponse.

I made the change from JSON to file and when I ran the automation I got this error.

Any ideas on what this issue means and how to fix it?



Are you usingn on prem orchestrator?

Or a company laptop looks like the url is blocked or there is acerrificate wrror

Can you opwn exception details from locals panel and check inner exception please


@Anil_G Looks like that was the problem. I got the URL whitelisted and the issue is gone.

Thanks for your help

1 Like

One more question about MLSkills


I’m getting this output from the JSON response in the MLSkills activity.

Any ideas on why the prediction is failling?


Looks like some issue with the file…but it says internal server error which occurs when url is down…can you check if the skill is disabled


Should there be an active pipeline on the project for this to work properly?

I’m not sure where to look to make sure that the skill isn’t diabled.

@samuel_trahan are you using document understanding framework?If you use that you can try one activity that is machine learning extractor.

In this activity you can indicate the url of your ml skill if it’s public or you can load it directly if you are in the same tenant. Then is going to load all settings of your ml skill.

To make sure your ml skill is active you can check it in action center. There you have the option to update not skill and also disable.

Let me share some useful links of documentation:

Also if you want to understand better how to use document understanding framework I highly recommend @Lahiru.Fernando tutorials on YouTube.

@Camila_Caldas Thanks for your response.

I was using machine learning extractor before I started to implement Machine Learning into my document understanding.

The process was very manual with the users having to validate every purchase order that comes through the process.

I set up the project with digitization, classifiers, extractors, validation station, and training. I’m trying to put the MLSkill and Upload File into those sections, and I’m curious if I should remove all of the steps that were there before meaning ( digitization, classifiers, etc.) and just use MLSkill and Upload file?

I will check out the youtube channel that you recommended.


This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.