Privacy & Data Handling
This one-pager summarizes Ambient's privacy and data handling posture for API inference traffic and optional chat features.
Core commitments
- Zero content logging / retention for inference: Ambient does not log, store, or retain inference request or response content (prompts/inputs, files/attachments, tool payloads, or outputs). Content is processed only transiently to generate and return the response.
- No training or secondary use: Ambient does not use customer Content or Outputs to train, fine-tune, evaluate, or improve models or services. Content is not used for advertising, profiling, or analytics.
- Optional Chat Data storage only at user request: Chat transcripts/history are stored only if a user chooses to save them in the interface. Users can delete chats in-product at any time. Bulk deletion (including account-level deletion) is available by request via [email protected].
Operational data (non-content)
Ambient may retain limited non-content operational data to operate and secure the service (e.g., authentication events, billing records, timestamps, latency metrics, error codes, rate-limit events, and aggregate usage totals). Ambient designs telemetry and logging to avoid storing inference Content.
Vendors
Ambient uses Cloudflare (security/network delivery), Stripe (payments), and Sentry (error monitoring) for operational purposes. Inference Content does not touch these vendors.
Third-party routers (e.g., OpenRouter)
If a third-party platform routes requests to Ambient, Ambient processes Content only to generate and return the response and does not retain it. The third-party platform may independently log or retain Content under its own policies.
For details, see the published Privacy Policy and Terms of Service.