AI Device Template Featuring Whisper, TTS, Groq, Llama3, OpenAI
github.com
github.com
One wrong permission and your pictures, contacts etc get uploaded.
So you only put on the device what you want to upload anyway.
The AI is always server-side. So the device is useless.
Whenever someone figures out how to package the GPU (or LPU, TPU, *PU) into a wearable or portable "box", that device will finally make sense.
... and it will most likely just be a smartphone.
So for something like this: It has to start with minimal hardware for voice recording, some local small computer just capable enough for simplest local NLP processing. A common open protocol for mesh communication and other local offline hardware to hardware features. Then it should decide when its necessary to reach out and communicate one to one, or one to a group, or send the command up to AI cloud processing because the lower layers cant do whats being asked.