July 9, 2026 · 4 min read
Text is a narrow way to talk to an AI. You describe a thing in words, the model imagines it, and something is always lost in translation. Scowld takes a different route: with the camera enabled, the companion can respond to what you physically show it — hold up an object, point at something on your desk, or just let it see the room.
This vision feature is optional and off by default. When you turn it on, it turns a one-way chat into something closer to showing a friend what you're looking at.
When vision is enabled and your message would benefit from it, Scowld captures a single frame from the front camera and sends it to the AI provider you configured alongside your text. The provider's vision model reads the image and the companion replies out loud.
Vision is opt-in and scoped to the moment you use it. There is no always-on recording, and Scowld does not run a server that sees your images — the frame goes straight from your device to the provider whose key you added.
Scowld also does not use Apple's TrueDepth API or ARKit face tracking, and it does not collect, store, or share face geometry, depth maps, facial expressions, or any other face data. The camera is there to see the world you point it at, not to profile you.
Download Scowld from the App Store, add an AI provider key that supports vision in Settings, then enable the camera from the composer and send a message. Because Scowld is bring-your-own-key, you only pay your provider for what you use — there's no subscription.
Does Scowld record video?
No. Vision captures a single still frame only when you send a message with the camera enabled. There is no continuous recording, and images are not saved to your photo library.
Which providers support vision in Scowld?
Image input works with vision-capable providers such as Google Gemini, OpenAI, and Anthropic Claude. You choose the provider and add your own key in Settings.
Scowld is a free, open-source iOS AI voice companion. Download it on the App Store or read more about Scowld.