I wish I had these models when we were building speaklish. Live avatars in any form, big thing edtech and any content creating
Many think, advancement of AI is only happening around chatbots or LLMs, but I believe Google(DeepMind) is one of such companies contribute to other fields of sconce equally, remember AlphaFold ? but unfortunately, they also recently shut down it to focus on more "AI"
So interestingly, those are considered frontier Labs if they work on exclusively on llm problems ? if you digger deep these labs trying to solve intelligence problems, if you solve this one problem it all the others, but question we can solve it? is not it better to so solve existing rather than spending resources for unlocking generality?
Maybe you also saw again recently openai claimed that they solved the Navier–Stokes million dollar math problem, it's under discussing that did really 10,000 of internal agents solve it? did humans bring all fundamentals and agents put them, together, but personally for me math is deep thinking fields, adding concurrency definitely boosts progress, because problem is well defined and output can be verified, while other fields like chemistry and cs, physics needs more experimentations, and thinking deep isnot enough.
Another wave AI agents is already here, basic function calling update of openai now enabled billions of dollar new economy for agents
Muse opeclaw, oai dots, grok bot now everyone is thriving for getting users and fighting for market share
Interesting times to be alive
Google
Introducing Gemini 3.8 Live with Live Avatar
Introducing Gemini 3.8 Live with Live Avatar, which brings near real-time visual presence to Gemini’s conversational AI.
1
1
1October 6, 2026 235 2
