Suprbox - Box for AI agents to secure enterprise data storage

Hey , I built Suprbox after watching a friend's team accidentally hand a research agent the same Google Drive credentials a human would use. Within a week the agent had read salary spreadsheets, board memos, and an un-redacted contract none of which it needed for its actual job. No prompt guardrail catches that; the agent was just doing what it was told, with the access it was given. So I started from the data side instead of the prompt side. Every agent gets its own scoped Suprbox key, every document lives in a vault with rules (sensitivity, rate limit, time-of-day, human approval, watermark), and every byte that leaves is signed and logged. If something goes wrong, you have a real audit trail not a chat log. The biggest shift while building was realizing this isn't really an "AI tool." It's storage with a policy gate in front of it. That reframing made the whole thing simpler the API ended up looking like S3, just with rules. Would love to hear how you're handling agent access today, and what feels broken about it. Happy to answer anything. — Hritvik