Recently updated channels

Pick one to see only that channelBrowse all channels →
20VC with Harry Stebbingsyesterday +46AI Engineeryesterday +1硅谷1012 days agoMBMatthew Berman3 days ago +57PragerU3 days agoY Combinator3 days agoa16z4 days ago +1SemiAnalysis4 days agoGOTO Conferences4 days agoThe Builders5 days ago

Results

1 articles
Sort
ChannelSequoia CapitalTopicAgent memory
44:52EN
Sequoia Capital· Jun 24, 2026PodcastAn 80GB Wikipedia Cache Versus a 100GB Model That Holds the InternetEngram’s founders argue that stuffing a company into the context window is the expensive kind of memory—and that baking it into the weights can cut token use by two orders of magnitude.RAG & knowledgeContinual learningFine-tuning+3