Privacy Policy
The short version: WDYT has no servers and no accounts. Speech is recognised on your device. The only network calls are the two you initiate — understanding your question with Google Gemini or OpenAI, and deciding with TypeSafe Jev via OpenRouter or TypeSafe directly — and both go straight from your phone to the provider you picked in Settings, with your own API keys.
What we collect
Nothing that identifies you. WDYT has no accounts, no analytics SDKs, no advertising, and no third-party trackers. We operate no server that your device talks to.
Microphone and speech recognition
When you record, your speech is transcribed on the device: English with Apple's Speech framework (depending on your device and iOS settings Apple may process speech on its servers under Apple's privacy policy), Mandarin with the SenseVoice model running locally via sherpa-onnx. WDYT does not store audio; the transcript exists only for the current question and you can edit it before anything is sent.
Photos
Photos you attach are resized on the device (longest edge 1280 px) and sent only to the generator you picked (Google Gemini or OpenAI) as part of understanding your question, using your own key. They are never sent to the decision model or anywhere else, and are discarded when you start a new question. WDYT does not access your photo library beyond the pictures you pick.
Your API keys and the two providers
WDYT works with two keys you paste in Settings, one per step, each for the provider you select there. Understanding the question: with a Gemini API key, your transcribed words, photos and profile (see below) go directly to Google under Google's privacy policy; if you turn on Grounding with Google Search, Gemini may also run web searches for the question. With an OpenAI API key, the same content goes directly to OpenAI under OpenAI's privacy policy. Deciding: with an OpenRouter API key, the confirmed question, options and profile go directly to OpenRouter, which routes them to TypeSafe's Jev model, under OpenRouter's and TypeSafe's policies; with a TypeSafe API key, they go directly to TypeSafe under TypeSafe's policies. Only the provider selected for each step receives anything. Keys are stored in your device's Keychain, never leave the device except to the provider they belong to, and are removed when you clear the field or delete the app. WDYT never receives your keys or your content.
Your profile: identity.md and soul.md
Two plain-text files on your device describe your preferences and how you decide. You can write, edit or clear them in Settings at any time. They are included in the requests above so answers fit you. If you turn on update these from my questions (off by default), after each answer your question and edits are sent to the generator you picked (Gemini or OpenAI) with your key, which returns revised files; nothing else is stored, and the setting can be turned off at any time.
Data stored on your device
Settings, the two profile files, your recent decisions and the downloaded SenseVoice model live on your device. The SenseVoice model is stored in a shared container so other apps from the same developer (tautiu.dev) can use the same copy. Deleting the app removes its data; the shared model remains until the last app using it is removed.
Children
WDYT is intended for users aged 13 and up and does not knowingly collect data from children. It collects no personal data from anyone.
Changes
If this policy changes, the new version is published at this address with a new date, and material changes are noted in the app's release notes.