Who processes your data

We share your information only with the providers that run the service for us, under contracts that limit them to that job. Here is each one in plain words, and where it is.

Behind AI Within

• Amazon Bedrock (United States) runs the language model. It receives your messages, voice transcripts, the account information AI Within reads and memory text. AWS does not train on it, and we keep model-invocation logging switched off, so Bedrock holds no copies.
• Deepgram (United States) turns your voice into text in a voice session. Every request opts out of its model-improvement programme, and audio is held only for as long as it takes to transcribe.
• Amazon Polly (United States) turns AI Within's replies into speech. Our AWS account opts out of all AI-service data use, so nothing is kept or used for improvement.
• Jina AI (Germany) indexes each remembered fact so AI Within can find it later. It does not train on your data.
• LangSmith (United States) is a debugging tool. It receives 1 in 20 ordinary conversations, kept for 14 days. Conversations about a crisis are never sent.
• Amazon CloudWatch (United States) holds application logs: identifiers, counts and error text, never message text, kept for 30 days.

Around your account

• Resend (United States) sends our emails. If you report an AI response, the email to legal@linkedwithin.com includes the first 2,000 characters of the reply you reported.
• Stripe (United States, global) handles website payments. Stripe holds your card details; we never see your full card number.
• RevenueCat, Apple and Google handle app-store subscriptions, sign-in with Google or Apple, and push notifications. We receive subscription events and a sign-in identifier, never your payment details.
• ipapi.co (United States) looks up your country from your IP address once, at sign-up. We store the country, not the address.

We use no analytics trackers, advertising pixels or data brokers. If a provider changes, we update our Privacy Policy first.