Privacy
Chompute helps companies control AI usage by using request metadata, budget policy, and routing decisions carefully.
Data used for AI usage control
Chompute needs usage signals to help companies understand and control AI spend. The exact data depends on how a team configures the control plane and endpoint.
- Account, workspace, team, project, and user identifiers.
- Tool, API key, model, provider, route, timestamp, latency, token, and cost metadata.
- Policy events such as alert, throttle, stop, fallback, or endpoint routing decisions.
- Operational diagnostics needed to troubleshoot failed or slow requests.
Prompt and response handling
AI gateways sit in the request path, so teams should choose retention and logging settings deliberately. Chompute messaging should stay focused on usage control unless a specific deployment has a written prompt-retention policy.
For sensitive environments, teams should minimize stored payload data, restrict access to logs, and keep policy decisions auditable.
Endpoint routing
The flat-rate endpoint is meant to reduce the cost of appropriate workloads. Routing rules should make clear when work is sent to Chompute Endpoint, when it stays with an existing provider, and when it is blocked by policy.
- Route routine or repeated work when flat-rate capacity is more economical.
- Keep restricted workloads on approved model paths.
- Avoid silently changing model behavior for high-risk workflows.
- Use budget, priority, and project context to make routing decisions explainable.
Access keys
User-specific API keys are private. Public Chompute pages should not contain user-specific API keys. Endpoint keys should be copied only from authenticated Chompute surfaces.
- Do not commit keys to GitHub.
- Do not paste keys into public chats or marketplace listings.
- Do not include keys in screenshots.
- If a key is exposed, rotate or revoke it if available, or contact Chompute support.
Admin access
AI usage and cost data can reveal sensitive information about internal projects and team behavior. Admin access should be limited to people who need it for finance, platform, security, or engineering operations.
- Use least-privilege access for dashboards and logs.
- Separate production and development usage where possible.
- Review exports before sharing usage data outside the company.
- Keep budget and routing policy changes visible to authorized admins.
Contact
For privacy questions about Chompute, contact support@chompute.ai or use the Chompute support page.
Keep controls close to the workflow.
Route AI traffic through documented control surfaces, keep access keys private, and make budget and data policy explicit before usage scales.