Understanding does not mean surveillance.
What vloop stores, what it never stores, and how to delete your data. Plain language, not legalese.
vloop is in the understanding business. Understanding does not mean surveillance.
vloop frames your prompt so your AI understands what you meant — then the selected model writes the answer. vloop returns structure only; it does not save prompt history or answer history. This is a plain-language summary of what is true today, not a final legal privacy policy.
The boundary
What vloop stores
- Account identity — your vloop sign-in account.
- Account entitlement — your private-alpha access status, plan, and monthly limit.
- Usage metadata — a monthly request count only.
- Connector token metadata — a one-way hash of your connector token, plus its channel and dates.
All content-free — counts and status, never your words.
What vloop never stores
- Your prompts
- The model’s answers
- Routes
- Frames
- Model output
- Hidden reasoning
- Conversation history
It is never written down — so there is nothing here to leak, sell, or hand over.
How connector tokens work
When you connect vloop to where you already use AI, vloop issues a connector token. The raw token is shown once and kept only as a one-way hash of your connector token — vloop cannot recover the original. You can revoke connector tokens any time in Billing; revoking disconnects that place immediately.
How deletion works in private alpha
To remove your data, revoke connector tokens in Billing and request deletion via vloop support. An operator removes your content-free records — entitlement, usage counters, and connector token hashes. Because vloop returns structure only and never stores prompts or answers, there is no conversation history to delete.
Developer details — storage flags
- Prompt storage: Off
- Answer storage: Off
- Internal intelligence storage: Off
- Motion storage: Off
- Telemetry: Off
- Account status: On
- Billing status: On
- Usage count: On
When you choose a live provider, your prompt is sent server-side to that provider so the answer can be generated — it is used for that response and not persisted.