Using Copilot to analyze SharePoint document libraries and content.

I am just getting started with Copilot studio. I've been asked to build a prototype where we use Copilot to analyze some large document libraries across several SharePoint sites. 

The short term goal of this project is to use CP to help us classify/categorize documents into clinical (medical) and non-clinical. For instance, I would want to ask CP to get me an excel list of documents with a column indicating if the document is clinical or not.

The long term goal is to use Copilot to apply data classification/sensitivty labels to the documents if they are clinical. 

We have several sites with libraries where there are multiple TBs of documents that are intermingled (clinical and nonclinical). This could really help us organize this information. By doing this, it gets us in a good position in regards to our data retention policies.

I have Copilot studio connected to SP with a delegated app registration. I can send it a name of a document and it can tell me if it contains clinical data or not. I want to scale this up so it do this at a larger scale.

My questions are:  Is this even possible? If so...how would I approach this?  In studio, when I ask basic questions like "How many documents are in this SharePoint site", it says it can't undestand and to rephrase.  However, I can ask: Give me the newest document that was created...and it will return a correct result.

Any help or direction for me would be huge!

Thanks!