What should a trustworthy human writing certificate actually prove?

We recently launched Hummint on Product Hunt.

The core idea is simple: instead of trying to guess whether a final text “looks AIgenerated”, Hummint creates verifiable evidence around how the text was written.

We’re building a proof layer for human writing: writing sessions, revision signals, timestamps, hashes, AI disclosure, and a shareable certificate.

I’d love feedback from students, professors, journalists, editors and makers:

What would make a human writing certificate genuinely trustworthy?

• Writing and revision history?

• Timestamped checkpoints?

• Cryptographic verification?

• Privacy preserving proof?

• AI use disclosure?

• Something else?

Critical feedback is more useful than compliments right now.