Maple is a great step toward privacy, similar to GDPR.
The models themselves can handle large amounts of data when pasted into the box. Especially DeepSeek and gpt-oss.
We do plan to bring back document upload in the future, it is a highly requested item. The system is getting an architecture upgrade so it can handle that better.
One option is the Maple Proxy. We have users who have their own document solution locally using another app (Goose, for example) and need Maple for the encrypted LLM as the final step.