Sometimes you want the companion to react to what's actually in front of you, not just what you type.
Scowld has an optional vision feature. When you enable it, front-camera context can be sent to the AI along with your message, so the model can factor in what it sees when it replies.
A few honest notes on how this works:
- It's optional and off until you turn it on. Vision context only gets sent when you've enabled it.
- It uses the front camera, the one facing you, so the companion can have visual context for the conversation.
- Because Scowld is bring-your-own-key, the vision context goes to whichever provider you've configured for the LLM. You choose the model that's doing the looking.
Combined with voice and the animated VRM character, vision makes the interaction feel a lot more present. You talk, it sees, it answers back in its own voice with a face that reacts.
The app is iOS only and built solo, MIT licensed. If you've been curious about wiring camera frames into a multimodal chat loop, the source is open.
Start from the site.
Top comments (0)