The main reason I fell in love with tech as a kid was cause it was the closest thing to magic in real life. With LLMs we can now talk to machines, but through old generation of interfaces. I'm convinced that AI should be wearable and always on. That we will be able to talk to AI at any second in any setting without touching a smartphone. Thus we need some way of talking to it as we would to a college sitting next to us, but without actually saying stuff out loud (you rarely use Siri/subway in office). The technology of reading human speech without us actually making a sound has been around for years and is not new. But usually it needs to read data directly from muscles with electrodes on your face, which is not an everyday wearable and quite frankly doesn't make you much hotter. Essentially the challenge is how to find a smart way to shrink all the stuff on video into easy to wear every day device. That's the idea behind @engram_labs. On the vid, Im recording data for training a teacher model for silent speech, in case you're into hardware and stuff - let's chat! I'm looking for a co-founder and also need more data for the training.

Read the post on X