Hello! What an interesting use case...Hmmm…
I’m not actually sure, but in principle, what you’re trying do should be straightforward if you’ve already set up your algorithms and can access them via an API.
The workflow would be:
- Document Trigger
- Send document to your service for categorisation and analysis
- Service sends return value with categorisation data
...Although if you’re asking for a Machine Learning service recommendation, I’m sorry, I’m not really the best person to ask
@Paolon - Sounds like something more out of a document/information management (or even eDiscovery) use case than what we’ve traditionally done with Zapier. Getting the document to your system should be relatively simple, depending how it’s set up to intake information. My one concern there would be scale. Zapier is ultimately based on individual tasks, while we’ve worked with engines that are meant to (and need to) ingest gigabytes of documentation every hour.
If “classification” here is simply, say, routing to the right Google Drive folder, you’d be in luck – so long as your system can take it from there and post back somehow.