筛选结果
29 篇排序
1:35:29ENUber Burned a Year of Tokens in 13 Weeks. Retrieval Is How Agents Win主持 Nathan LabenzMongoDB Field CTO of AI Pete Johnson argues that agent performance now hinges on retrieval quality, not on stuffing million-token context windows.
2:38:48EN → 简体中文一千次搜索只要五分钱,高分也不必靠作弊主持 Nathan Labenz在《Cognitive Revolution》的一期早间节目里,Ceramic 的 Anna Patterson 把廉价搜索、GPT-5.5 干净的 Vending-Bench 成绩、Zvi 关于模型福利的主张,以及能把矩阵乘法能耗降低约 30 倍的模拟芯片放在一起讨论。
1:40:51ENReasoning Models Just Made the Other 80–90% of Enterprise Data Usable主持 Nathan LabenzSnowflake VP of AI Baris Gultekin on text-to-SQL, RAG, agent design, and why inference has to run next to governed data.
1:41:34ENChat Is the Universal Interface—and Often the Worst One主持 Nathan LabenzAtlassian's head of AI on fighting slop with taste, knowledge, and workflow, and why chat is a terminal, not a product.
1:55:12EN → 简体中文28,000 份文献、一个 30 亿参数的模型、22 个人:天主教 AI 的技术账本主持 Nathan LabenzLongbeard 首席执行官 Matthew Harvey Sanders 在 The Cognitive Revolution 节目中讲述,一支使命驱动的团队如何让模型与天主教训导对齐,以及为什么微调还不够用。
1:37:25ENFin Climbed From 35% to 65% Resolution. Intercom Says the Models Only Did a Few Points主持 Nathan LabenzEoghan McCabe and Fergal Reid on production A/B tests, 99-cent outcome pricing, and why GPT-4 was already smart enough for most customer service.
1:25:12ENHe Hasn’t Typed a Line of Code in Months. Here’s How Augment Reads a Million-Line Repo主持 Nathan LabenzGuy Gur-Ari, co-founder and chief scientist at Augment, tells Nathan Labenz how coding agents actually work inside messy enterprise codebases.
56:07ENBox Holds 100 Billion Files. Its CEO Says Enterprises Only Move for a 10x主持 Nathan LabenzAaron Levie tells Nathan Labenz why AI excitement in the enterprise already outruns the early cloud — and why 98% reliability, seat-based SaaS, and incremental savings still will not ship.
1:29:00EN14,000 people applied. Thirty-three got in. The jobs are at Homeland Security.主持 Nathan LabenzMichael Boyce, director of the DHS AI Corps, tells Nathan Labenz how the largest civilian AI team in the U.S. government is being built — and what it is already shipping.
1:55:53EN → 简体中文AI 读论文已超过人类,却仍数不清原子主持 Nathan LabenzFuture House 科学负责人 Andrew White 梳理了语言模型在化学和生物学中已经能做什么,以及它们无法消除的瓶颈。
55:16ENGemini’s API Grew 14× in Six Months. Google’s PMs Explain the Catch-Up.主持 Nathan LabenzLogan Kilpatrick and Shrestha Basu Mallick walk through search grounding, Gemini Flash’s price-latency quadrant, and why a near-tied leaderboard still leaves Google third in most developers’ queues.
2:03:55EN → 简体中文约束越多,智能体(Agent)越值得信任主持 Nathan LabenzAgent.ai 工程负责人 Andrei Oprisan 向 Nathan Labenz 解释,为什么狭窄、规定明确的工作流能胜过开放式的自主性,以及一个 12 人的团队如何在几个月内把一个免费的智能体网络做到超过 40000 名用户。
1:51:00ENStart With Ten Gold Examples: A Three-Step Playbook for AI Automation主持 Nathan LabenzThe day after OpenAI opened GPT-4o fine-tuning, Nathan Labenz unpacks a three-year method for choosing work, extracting tacit judgment, and raising model performance.
1:21:53ENHippoRAG Matches Iterative Retrieval at One-Tenth the Online Cost主持 Nathan LabenzA hippocampus-inspired knowledge graph lets multi-hop questions skip the brute-force document crawl.
1:27:01ENThree Million Emails, Ten LLM Calls, Six Models: Shortwave's RAG Playbook主持 Nathan LabenzFirebase founder Andrew Lee walks Nathan Labenz through the retrieval stack that turned a Gmail inbox into a working AI assistant.
1:23:09ENThe Right Size of an AI Task Is Whatever You Can Check主持 Nathan LabenzElicit founders Andreas Stuhlmüller and Jungwon Byun walk through how a twelve-person research assistant cut summary hallucinations from 1.5 to 0.5, ships about every week and a half, and treats task decomposition as the way to spend more compute without a black box.
1:28:37ENTwelve Hours After a GPT-4 Demo, He Turned a Decade-Old Legal Startup Around主持 Nathan LabenzPablo Arredondo, co-founder of CaseText and now VP of CoCounsel at Thomson Reuters, traces legal research from 19th-century citators to a retrieval-anchored GPT-4 product that law firms treat more like a colleague than a search box.
Google Gives You 33 Million Results. Exa Wants 54.主持 Nathan LabenzWill Bryk, CEO of Exa, tells Nathan Labenz why search and research are different jobs, and why an AI librarian for the web has to be neural from the first filter.
1:32:19EN → 简体中文2018 年的提示词论文被拒,如今他的搜索 API 超过了谷歌主持 Nathan LabenzRichard Socher 讲述 You.com 的来路:从一个被拒的统一自然语言处理(NLP)模型,到原生于大语言模型(LLM)的网页索引、三种助手模式,以及他给出的末日概率(P(doom))非常低。
1:53:08ENWhen GPT-4 Writes SQL, Four Tables Is the Breaking Point主持 Nathan LabenzJuan Sequeda on why a knowledge graph is the accurate, explainable brain that foundation models do not have.
1:50:57ENTwenty-Eight Days After GPT-4, an AI Designed and Ran a Chemistry Experiment主持 Nathan LabenzGabe Gomes and his Carnegie Mellon students built Coscientist, a language-model agent that plans, codes, and executes real experiments on a remote cloud lab.
1:32:40ENThe Historian Who Taught GPT-3.5 to Beat GPT-4主持 Nathan LabenzMark Humphries on a kilometer of unread paper, fur-trade jargon that embeddings cannot see, and why cheap processing will force history to be rewritten.
1:26:30EN → 简体中文这些数据大多从未进过数据库:Chroma 战时 CTO 谈 RAG主持 Nathan LabenzAnton Troynikov 重返《认知革命》,主张检索质量主要取决于嵌入(embedding),Chroma 应当掌控检索增强生成(RAG)的整个闭环;GPT 则仍然通不过一项基础推理测试:康威的“生命游戏”(Game of Life)。
1:19:52EN1,000 H100s Take 115 Days; 22,000 Take Five主持 Nathan LabenzNathan Labenz and Trey Kollmer walk through cluster math that turns GPT-4-scale training into a weekly experiment, then four papers on search-augmented models, recursive code improvers, spatial maps inside Llama 2, and what networks do when they have no idea.
加载更多已显示 24 篇,共 29 篇









