Training superhuman coding models at Cursor

Cursor's Sam Whitmore sits down with Baseten's Charlie O'Neill and Harry Partridge to discuss where working with agents goes next: running them by the hundred, getting them to message and review each other, and building systems instead of just managing parallel tasks.

Sualeh Asif (Co-Founder of Cursor) sits down with Simon Eskildsen (CEO of turbopuffer) to discuss scaling Shopify through flash sales, why a new database company emerges every 15 years, and what makes a P99 engineer.0:00 - Scaling Shopify through flash sales and outages7:11 - How the Shopify infrastructure team evolved8:13 - How top infrastructure teams collaborated in the 2010s10:35 - Engineering principles from Logrus and on-call17:38 - The story behind Simon’s famous-ish blog, Napkin Math23:05 - Why new database companies keep winning32:21 - How Simon became a fan of databases35:45 - AI coding, and where agents still fail38:44 - What it would take for turbopuffer to become Google-scale42:10 - Hiring P99 engineers in the AI era48:45 - What's next for databases

A conversation with John Schulman on the first year LLMs could have been useful, building research teams, and where RL goes from here.00:00 - Speedrunning ChatGPT09:22 - Archetypes of research managers11:56 - Was OpenAI inspired by Bell Labs?16:54 - The absence of value functions18:23 - Continual learning21:09 - Brittle generalization24:05 - Co-training generators and verifiers, GANs27:06 - John’s personal use of AI for research28:54 - Day in the life33:01 - Slowdowns in consequential ML ideas36:21 - "Peer review" within the labs39:19 - Distribution shift in researchers43:33 - Future of RL45:33 - Will the labs coordinate if the world needs them to?44:46 - Forecasting ills in AGI and engineering47:53 - Thinking Machines

Michael Truell (CEO of Cursor) sits down with Patrick Collison (CEO of Stripe) to discuss programming languages, the role of AI in programming, and building long-lasting software.00:15 Why Patrick wrote his first startup in Smalltalk03:35 LISP chatbots06:09 Good ideas from esoteric programming languages09:12 Brett Victor and Dynamicland16:37 Programming human organizations20:28 A codebase's "Big Bang" moment and MongoDB25:49 Rewriting Stripe32:00 How do you, Patrick Collison, use AI?38:25 Changes to GDP/TFP41:56 Programming human biology46:10 Unexpected beneficiaries of AI

What does it really take to train an AI to write great code? In this roundtable conversation, some of the founders and researchers at Cursor explore the process of building superhuman AI coding agents — from infinite context models and the best rewards to real-time RL and scaling to millions of environments.This is an inside look at some of the work behind Cursor and what AI-assisted coding might look like in the future.