PERSON DIRECTORY
Jeffrey Quesnelle
Jeffrey Quesnelle appears in 2 indexed conversations across Practical AI, Sina Habibian Podcast. This directory brings every appearance, source, TL;DR, digest, and transcript into one searchable feed.
Hermes Agent: Agents that grow with you |Episode #357|
Daniel WhitenackChris BensonJeffrey Quesnelle
Hermes Agent turns successful tasks into reusable skills, making a restaurant booking that took 30 or 45 minutes initially instant the next day.Nous’s debugging agent accumulated skills over about a month, showing how agent infrastructure can spread expertise and potentially sell cheaper or faster outcomes, while ownership and value capture remain unsettled.NVIDIA could sponsor Western open models because workloads run on its chips, though Quesnelle’s optimism is limited to the short term.
Jeffrey Quesnelle on Nous Research, large language models, and the human mind
Sina HabibianJeffrey Quesnelle
Nous Research’s bet is that open AI needs the capacity to train foundation models independently, with Psyche coordinating distributed GPU participation.Quesnelle says DisTrO cuts training communication requirements by roughly 1,000:1, though each node still needs substantial fast memory and local interconnect.The approach could make frontier research less dependent on a few model publishers, while verification bottlenecks and unresolved safety risks remain.