Training Data cover art
Podcast · 110 episodes

Training Data

by Sequoia Capital · English

Join us as we train our neural nets on the theme of the century: AI. Sonya Huang, Pat Grady and more Sequoia Capital partners host conversations with leading AI builders and researchers to ask critical questions and develop a deeper understanding of the evolving technologies—and their implications for technology, business and society. The content of this podcast does not constitute investment advice, an offer to provide investment advisory services, or an offer to sell or solicitation of an offer to buy an interest in any investment fund.

Latest episodes

All episodes

Box's Aaron Levie: On Reinventing Yourself in the AI Age and Enterprise DiffusionReinventing Box/enterprise software for the AI era by focusing on the “bridge” between LLM capabilities and real enterprise workflows; why application-layer diffusion will be massive; how agents…15 Sep 2026 · 1 h 5 min · 31 chapters
Making Cities Awesome: Peregrine’s Nick Noone & Ben RudolphPeregrine’s approach to using AI to improve city and public-safety outcomes while rejecting a surveillance-state model.1 Sep 2026 · 52 min · 18 chapters
Parallel’s Parag Agrawal: Building a New Web for AI AgentsParallel Web Systems is rebuilding web search for agentic AI, arguing that human click data is a “bug.” Instead, agents should provide feedback to improve indexing/ranking, using model research to…25 Aug 2026 · 55 min · 24 chapters
Rich Sutton and Khurram Javed: Why AI Models Stop Learning, and How to Start It AgainWhy AI systems stop learning (weights frozen after pre/post-training), and how to enable continual deep learning plus continual world-model updates and abstraction/planning.18 Aug 2026 · 54 min · 26 chapters
Chai Discovery's Bitter Lesson: Drug Design Is Another Scaling ProblemChai Discovery co-founders Josh and Matt argue drug design is a scaling problem and should be made more “engineering-like” via de novo molecule generation, rigorous evals, and simplification of model…4 Aug 2026 · 47 min · 25 chapters
Building the Automated AGI Lab: Core Automation's Jerry Tworek and Rohan AnilCore Automation’s founders argue that replacing Transformers requires (1) architectures that learn at test time/continually adapt, and (2) end-to-end integration of pre-training + reinforcement…29 Jul 2026 · 49 min · 20 chapters
Factory's Matan Grinberg: The Coming ‘Dark Factory’ Where Software Builds ItselfFactory’s co-founder Matan Grinberg discusses “dark factory” software factories—autonomous agents that build software—plus why enterprises need model independence, modularity, and cost-aware routing…21 Jul 2026 · 52 min · 18 chapters
Anthropic's Katelyn Lesse & Angela Jiang: Building an Ecosystem, not a Walled GardenAnthropic platform strategy for building an open ecosystem for Claude-based agents, using a “layer cake” of abstractions (knowledge, execution, coordination) and moving from token-heavy…14 Jul 2026 · 49 min · 23 chapters
Inside Zipline's Autonomous System: 140M Miles, Zero IncidentsZipline’s autonomous logistics system for life-saving deliveries, focusing on why the drone is only a small part of the solution, and how they achieve safety, reliability, precision, and scale (140M+…7 Jul 2026 · 55 min · 30 chapters
Why Hardware-Software Co-Design Is AI's Real 100x: Dylan Patel of SemiAnalysisHardware-software co-design as the main driver of AI efficiency (the “real 100x”), with a focus on inference benchmarking, cost/throughput tradeoffs, and where bottlenecks will be next (memory…30 Jun 2026 · 1 h 10 min · 31 chapters
Memory and Continual Learning: Engram's Dan Biderman and Jessy LinEngram co-founders Dan Biderman and Jessy Lin discuss memory and continual learning for AI systems that must adapt to new, private, team-specific context.24 Jun 2026 · 45 min · 20 chapters
Simulating Humans at Scale: Simile's Joon Sung ParkJoon Sung Park, founder/CEO of Simile, explains building “applied AI” simulations of human behavior and societies to test decisions at scale, grounded in real behavioral data.16 Jun 2026 · 39 min · 15 chapters
Google DeepMind's Logan Kilpatrick: Why the Model Eats the HarnessGoogle’s agentic AI push—Gemini “agent harness” (Antigravity), agentic coding, and Omni world-model video editing; how this changes product strategy, cannibalization fears, and startup opportunity.11 Jun 2026 · 51 min · 26 chapters
LIVE: Jensen Huang on Building the Dynamo of the Intelligence AgeJensen Huang explains the “AI factory” concept and argues AI is the next major infrastructure buildout, comparable to the energy and internet revolutions.10 Jun 2026 · 41 min · 13 chapters
Knowing What Your Customers Want, All the Time: Listen Labs' Alfred WahlforssListen Labs’ AI-first customer research platform scales market research by running thousands of voice interviews simultaneously, building searchable audience profiles, and eventually simulating how…2 Jun 2026 · 42 min · 29 chapters
How Cursor Trained Composer on Fireworks: Distributed Infrastructure for High-Performance RLCursor’s Composer2 training and the distributed RL infrastructure behind it, including why environments must closely mimic real user computers to prevent “cheating,” plus systems tricks to run…26 May 2026 · 46 min · 21 chapters
Rebuilding IT From the Ground Up for the AI Age: Serval's Jake StauchServal’s AI-native enterprise service management for employee support, aiming to “close the gap” between what employees expect from IT help and what they actually experience.19 May 2026 · 38 min · 16 chapters
Suno's Mikey Shulman: Everyone Can Make Music NowMikey Shulman (Suno) explains how Suno turns text prompts into full songs by modeling audio directly as sound waves, why Suno treats music as a “creative entertainment” creator platform, and how AI…13 May 2026 · 36 min · 20 chapters
ElevenLabs' Mati Staniszewski: How Voice Becomes the Interface for EverythingMati Staniszewski (co-founder, ElevenLabs) explains how voice becomes an interface for dubbing, voice agents, and music; why ElevenLabs built audio “frontier” models without huge compute; and what’s…8 May 2026 · 27 min · 12 chapters
Anthropic's Boris Cherny: Coding's Printing Press MomentAnthropic’s Claude Code (Quad Code) and agentic coding—how it was built, why it “worked” once models improved, and what software development and startups will look like as AI makes coding…5 May 2026 · 25 min · 11 chapters
Waymo's Dmitri Dolgov: 20 Million Rides and the Road to Full AutonomyDmitri Dolgov (Waymo) recounts the path from DARPA autonomy to Waymo’s full autonomy, covering persistence through AI hype cycles, the Waymo Foundation Model (driver/simulator/critic…4 May 2026 · 27 min · 14 chapters
OpenAI's Greg Brockman: Why Human Attention Is the New BottleneckGreg Brockman (OpenAI) discusses why compute is still scarce, how scaling laws and architecture innovation drive model progress, and why human attention becomes the key bottleneck for agentic AI.1 May 2026 · 28 min · 13 chapters
Demis Hassabis on Building DeepMind, AlphaFold, and the Final Stretch to AGIDemis Hassabis discusses his path from chess and game AI to founding DeepMind, and argues for a “final stretch” toward AGI via reinforcement learning + deep learning, then using AGI-like systems as…30 Apr 2026 · 27 min · 12 chapters
Andrej Karpathy: From Vibe Coding to Agentic EngineeringAndrej Karpathy explains the shift from “vibe coding” to “agentic engineering,” arguing LLMs are a new computing paradigm (“software 3.0”) where prompting/context drives an interpreter, and progress…30 Apr 2026 · 30 min · 12 chapters