Y Combinator
MurmurCast publishes AI-generated summaries of Y Combinator’s YouTube episodes — 105 summarized so far, covering Scaling societies and social ratios, Comparative intellectual capacity across generations, Relative good versus bad behavior in scaled systems, Hierarchy and inferiority in advanced civilizations, Generational progress and advancement, AI capability timelines and superhuman AI development. Each summary distills the key insights, topics, and takeaways so you can decide what’s worth your time before pressing play.
We’ll Be Inferior to Our Grandchildren—and That’s Great
A discussion about scaling societies and the relative ratio of good versus bad behavior, with one participant expressing concern that future generations will be intellectually superior to current humans, similar to how humans are vastly superior to ants.
YC Debates: Should We Slow Down AI?
Jacob Coxin argues governments should substantially intervene to slow frontier AI development, citing rapid capability growth (superhuman AI researchers by 2027), alignment risks, and misaligned corporate incentives. Robin Hanson counters that typical technological adoption patterns suggest slower real-world impact, that regulation is dangerous without concrete problems, and that liability-based approaches are preferable to centralized global regulation.
How To Build The Future: Eric Glyman, Co-CEO of Ramp
Eric Glyman, co-CEO of Ramp, discusses how the company evolved from Parabus to build a $60 billion financial operations platform, and how Ramp is leveraging AI agents and internal tools like Inspect, Research, and Glass to transform organizational productivity and reshape the future of work.
Grok Bot: What Changes When You Can Build AI Teammates
Lauren Tan and Roman Ugarte discuss Grokbot, a platform for creating AI teammate agents with their own identities and tool access that can work autonomously. The conversation covers how AI agents are transforming software engineering and non-technical work, with Lauren showcasing her 4,000+ PRs/month workflow using specialized bots in Slack and describing the shift from execution-constrained to execution-abundant organizations.
Grok Bot: What Changes When You Can Build AI Teammates
Lauren Tan and Roman Ugarte discuss Grockbot, a product enabling users to create AI teammates with their own identities and capabilities. They share how AI agents are transforming software engineering and business operations, demonstrating workflows where engineers ship thousands of PRs monthly by leveraging specialized bots integrated into team communication tools.
Is AI Now Good Enough To Cure Cancer? | YC Paper Club
The YC AI for Bio Paper Club event explores how artificial intelligence is transforming cancer research and biological discovery through foundation models, personalized genomic analysis, and autonomous research systems. Speakers demonstrate that cancer is becoming a data problem solvable through AI agents, DNA language models like EVO2 are enabling efficient protein and genome design through few-shot prompting, and validated lab results show these approaches are moving from theory into clinical practice.
How Amjad Masad Saw Vibe Coding A Decade Early
Amjad Masad, founder of Replit, discusses how he predicted and built 'vibe coding' (AI-generated programming) a decade early by observing exponential technological trends and maintaining a culture of watching users. He explains that Replit's deep infrastructure investments in sandboxing, deployment, and collaborative systems positioned it perfectly to become the platform for AI-driven code generation and the emerging 'self-driving company' model.
Bringing Supersonic Flight To The People
Blake Schol of Boom Supersonic discusses how a failed partnership with Rolls-Royce forced the company to develop its own jet engine, which unexpectedly solved the sonic boom problem and created a highly profitable power turbine business for AI data centers. This pivot from paycheck-to-paycheck survival to capital abundance enabled Boom to optimize both airplane and engine design together, ultimately improving the passenger experience.
How To Go From Researcher to Founder | Startup School
A founder who was previously a researcher in academia outlines 10 critical mindset shifts required for academics to become successful startup founders, emphasizing the importance of customer validation, speed over perfection, and execution over novelty.
Engineers Make Great Founders
Two engineers, Shiobani and Chris Marcus, discuss their experiences at SpaceX and their decisions to start AI companies (Layers and Boot Loop). They highlight how SpaceX's unique engineering culture—where engineers owned complete product problems without product managers—prepared them to become founders.
Fixing AI Coding's Bottleneck
Blacksmith is a CI/CD infrastructure company that accelerated software validation by using gaming CPUs and custom cloud infrastructure optimized for CI workloads. Founded by engineers from Waterloo who identified that CI was universally slow across their previous employers, the company raised $66M at a $1B valuation and now serves customers like Langchain, reducing CI wait times from 15-20 minutes to 10-15 seconds.
YC Blacksmith v4
Blacksmith is a Series C-funded CI/CD infrastructure company ($66M at $1B valuation) that achieved 20x growth by optimizing continuous integration for modern development workflows. The company uses unconventional gaming CPUs and cloud architecture designed specifically for CI workloads, enabling dramatically faster build times and positioning itself as the solution for developer velocity in the era of agentic coding tools.
Why Everyone Is About To Have Their Own Database
Paul Copplestone, CEO of Supabase, discusses how the platform has evolved from a dashboard company to an API-first company and now to a code-first company optimized for AI agents. With 13 million users and 1 million databases created weekly, Supabase is experiencing explosive growth driven by both software engineers and non-technical builders who can now create sophisticated applications with AI assistance.
AI for the Air-Gapped Economy
A venture capital pitch highlighting the market opportunity in bringing frontier AI models to air-gapped, on-premises environments for security-sensitive industries. The speaker argues that the largest budgets and most defensible companies exist in this space, despite longer sales cycles and greater technical complexity.
AI for City Hall
AI can automate repetitive city permitting processes that currently cause significant delays in building and business launches. If U.S. cities approved permits 10x faster using AI, they would generate an additional $16 billion annually in property tax revenue, and contrary to conventional wisdom, selling software to governments is now increasingly viable.
LLMs That Write Well
LLMs have improved at coding and math but declined in writing quality, producing more robotic output than earlier models. This tradeoff stems from structural reward signals that prioritize technical tasks, creating a market opportunity for startups focused on conversational quality.
Vertical Foundation Models
The speaker advocates for vertical foundation models—specialized AI models built for specific domains like manufacturing, energy, and biology—that can match or exceed general-purpose models while requiring significantly less compute. These domain-specific models can learn new tasks with zero or few examples, with Google's Times FM3 for time series forecasting cited as a successful example.
Wall Street for the Physical Buildout
Physical AI infrastructure (robots, GPU clusters, data centers) requires massive upfront capital that traditional financing doesn't adequately address. The speaker argues this financing gap presents an opportunity for founders to build financial instruments and services tailored to the physical AI economy, similar to how railroads and electricity spawned new financial institutions.
Own Your Intelligence
The transcript discusses a shift in AI adoption from renting frontier models to companies owning and customizing open-weight models that are now competitive with proprietary alternatives. As open models improve and businesses become aware of their dependency risks, especially in regulated industries, companies increasingly want to own specialized models trained on their own data, run on their infrastructure, and continuously improve them.
Plaid for Government Benefits
Plaid proposes a digital infrastructure layer for government benefits that would verify income, tax, and residency data once and share it across all programs, reducing fraud and improving approval speed. The solution addresses systemic inefficiencies where applicants must repeatedly prove eligibility to disconnected government agencies using outdated manual verification processes.