About SaaufAI
SaaufAI is one gateway to models from many AI companies, with a plain YES/NO disclosure sheet for every model, and Wisps: assistants that keep working while you're away and ask before they act.
Who runs it
SaaufAI is run by a private individual based in Poland, not by a company. There are no employees or contractors: the operator builds the service, runs it and answers support. The operator's name is given on request by email to ai@saauf.xyz. The Legal notice has the details.
What it does
- The model gateway is one OpenAI-compatible API (
/v1/chat/completionsand/v1/models) to models from OpenAI-compatible hosts, Anthropic and Google Gemini, with streaming, tools, images and thinking levels. See the API reference and the OpenAPI spec. - Disclosures: every model is checked against the same questions about training use, retention, data location and capabilities, answered Yes, No or Not specified. Compare them on Compare.
- API keys can be locked to one model or provider, rate-limited, given a spending cap or an expiry date. Only a hash of each key is kept.
- Safeguards check requests against rate, token-budget, input-size and content rules before they reach a model, with fallbacks disclosed on each model's page.
- Wisps are always-on assistants with a name, a memory and their own email address. They run errands, answer mail and use the apps you connect, and ask a plain Yes or No before sending or changing anything.
Your data
The text of API and Playground prompts and answers is not stored: it passes through in memory to the model provider. SaaufAI keeps operational metadata such as the time, the model, token counts and status codes, and a short excerpt only when a content scan flags a Playground message for a person to review. Wisps keep their conversations and memories by design, until you reset or delete them. Nothing is sold or shared for advertising. The Privacy Policy has the full picture.
Where it runs
The website, API and Wisps run on Vercel; the database is on Supabase in the EU (Frankfurt). SaaufAI has no servers or offices of its own. How the service is protected is in the Security Policy. SaaufAI has signed the Cloud Security Alliance AI Trustworthy Pledge and publishes a STAR for AI Level 1 self-assessment, which is its own assessment, not an independent audit.
Contact
- Support, signed in or with your email address.
- Email ai@saauf.xyz for support, legal, privacy and abuse reports.
- Security problems, in English or Polish: see Security disclosure.
The policies are published in English and Polish.