[jhoetter]

Hi Tom! Thanks, happy to hear that :)

We've focused on JSON as the user-specified data model. So you can upload anything fitting into a JSON. We're using pandas to process the uploaded data, so spreadsheets or CSV-ish also work.

We've got a public roadmap (https://github.com/code-kern-ai/refinery/projects/1), and we're looking forward to also integrate e.g. native PDF labeling sometime soon.