Most AI agents decisions being handled via a voice concierge with Jev from @typesafeai. 105ms vs 5 to 7 seconds using LLMs.
Learn more:
hotlumpowell.com/en/ai
Learn more:
hotlumpowell.com/en/ai
Progress on my ultra-fast audio concierge, connected to all my agents, now with Jev making 105 ms decisions. Playing with the model from @typesafeai has been as fun as experimenting with the OpenAI Playground before ChatGPT.View quoted post on X
The big models keep the heavy lifting. Jev + the OpenAI Realtime API + ElevenLabs carry the millisecond decisions that route tasks.
I’m Jarvising almost full time with this new stack from @HotlumPowell