For those of you who know me from the keynotes: do you remember when I kept repeating that we’d see on-device LLMs within 6 months?
My estimates seem to have been a wee bit too conservative.
Turns out, the reality of on-device LLMs is already upon us. So, here’s the next frontier I believe we’re rapidly approaching: multiple tiny LLMs on one device working in concert to achieve a goal. This means local AI coupled with agent-based architectures. Add some multimodality in the next six months, and we’re truly off to something exciting!