Security built into every request
Rixi sits between your product and the AI models, so it is built to protect your keys and your users’ data from the first request.
sk_live_••••3f9aYou decide what Rixi keeps
Rixi keeps what billing and analytics need. Prompt and response text stays out unless you turn logging on.
- Never used to train models, by Rixi or by the model providers it routes to
- Encrypted in transit and at rest
- Deleted when you delete a user, from the Console or the API
Control who and what reaches your account
Secret keys never reach the browser, and each person on your team gets only the access their role needs.
Secret keys stay on your server
Secret keys only work from your backend. Rotate one in the Console without downtime.
Short-lived tokens for components
Embedded components get a token for one signed-in user that expires on its own, so the browser never holds your secret key.
Roles for your team
Owner, admin, developer and finance roles decide who can see keys, change limits or view revenue.
An audit log of every change
See who changed a key, a limit or a credit price, and when.
Card details never touch Rixi
When your users buy credits, they pay through your own Stripe account. Rixi only receives the result and adds the credits.
Checks, stores and charges the card
Card data stays hereLimits that stop abuse before it costs you
Per-user limits
Each user stops at their allowance or balance, so one account can’t run up your bill.
Rate limits
Requests are limited per user and per key, which slows down scripts and stolen keys.
Spend alerts
You hear from Rixi before spend passes the threshold you set, not after the invoice.
Ready for your security review
These are available on request during early access. Send us your questionnaire and we fill it in.
Security, answered
Are models trained on my users’ data?
No. Rixi doesn’t train models on your data, and it only routes requests to model providers whose API terms exclude training on that traffic.
Does Rixi store prompts and responses?
Not by default. Rixi keeps the request metadata billing and analytics need: user, feature, model, credits and time. You can turn on prompt logging for debugging and choose how long it is kept.
What happens if a secret key leaks?
Create a new key in the Console and revoke the old one. Until you do, per-user limits, rate limits and spend alerts limit what a leaked key can cost.
Can I delete a user’s data?
Yes. Delete a user from the Console or the API and their request history is removed. Purchase records are kept for as long as the law requires.
How do I report a vulnerability?
Email security@rixi.io with the details. Please give us a chance to fix the issue before sharing it publicly.
See the rest of Rixi
Bring us your security questions
We walk your team through keys, data and payments before you launch.