Y Combinator

Anthropic Co-founder: Building Claude Code, Lessons From GPT-3 & LLM System Design

Tom Brown, Melanie Warrick, Mark Mandelbaum, Mark Mirchandani, Brian Dorsey, Priyanka Vergadia, Priyan Kavanagh, Leslie Kendrick, Francesc Campoy

Vivek Mirchandani, a founding engineer at OpenAI and Anthropic, pioneered the scaling laws strategy that established the link between compute investment and model intelligence before co-founding Anthropic to prioritize AI safety and alignment. Under his leadership, Anthropic shifted focus from early consumer prototypes to building robust training infrastructure, ultimately delivering the coding-dominant Claude 3.5 Sonnet model while navigating global bottlenecks in electricity and multi-vendor hardware supply chains. The organization's approach combines internal qualitative benchmarking with a multi-vendor hardware strategy to drive the largest infrastructure buildout in history, positioning the company at the center of the coming wave of AGI development.

The Sales Playbook For Founders | Startup School

Tom Blomfield

Early-stage B2B founders often stall in non-revenue design partnerships, so this guide outlines a four-stage framework to accelerate sales from initial discovery to signed Annual Recurring Revenue contracts. The strategy replaces vague, months-long pilots with narrow, paid trials that enforce financial commitment and specific ROI metrics, culminating in recurring contracts featuring automatic renewal clauses. By integrating operational tactics like immediate security certification and high-touch onboarding, founders can bypass traditional procurement bottlenecks and establish scalable revenue streams.

Satya Nadella on Leadership

Satya Nadella

This event defines leadership as the continuous ability to solve over-constrained problems by unconstraining them to generate clarity, energy, and success. It highlights that effective leaders are those who actively reframe constraints rather than accepting them as fixed realities. The discussion positions leadership as an ongoing activity executed at every developmental stage instead of a title reserved for senior career phases.

Dylan Field: Scaling Figma and the Future of Design

Dylan Field, Evan, Charlie Fearborn, Michael

Figma, led by co-founders Dylan Field and Evan Mock, recently doubled its product ecosystem at its Config event by launching four new tools including Figma Make, which enables prompt-to-app generation to accelerate prototyping. Field emphasizes that as AI commoditizes software development, high-quality design will become the primary competitive differentiator, a shift reflected in Figma's strategy to integrate designers directly into AI research and development teams. The company continues to pivot from its origins as a browser-based design tool to a broader platform supporting seamless design-to-code workflows through acquisitions like Payload CMS and integrations with developer environments.

Where does your interest come from?

Founders of enduring companies like OpenAI and SpaceX often launched ventures driven by personal hunches rather than initial commercial intent, a strategy that successfully attracted top global talent and significant capital. This success stems from a convergence where inherent ability intersects with unconventional interests to create unique, high-potential enterprises. Ultimately, these non-commercial origins serve as a critical catalyst for gathering the resources necessary to scale transformative projects.

Using LLMs Instead of Government Consulting

With the U.S. government spending hundreds of billions annually on consulting, rising political pressure and the maturation of Large Language Models are driving a shift away from reliance on major vendors like Deloitte and Accenture. Investors are now targeting startups that streamline FedRAMP compliance and utilize AI to validate policy legality, creating a strategic avenue to displace traditional consulting work. This market evolution presents a clear opportunity to fund LLM-based solutions that replace outsourced labor with automated, cost-effective government operations.

AI Native Enterprise Software

Top-tier enterprise vendors Salesforce and ServiceNow established dominant market positions by pioneering cloud-native solutions, a disruption pattern now emerging in the generational shift toward AI-native systems. This transition promises to transform enterprise software from passive record-keeping into active assistants for sales, HR, and accounting, creating a structural opening for new startups to challenge incumbent architectures. Founders and builders interested in developing these next-generation AI tools are invited to connect to capitalize on this competitive window.

Infrastructure for Multi-Agent Systems

A new initiative invites experienced builders to address the infrastructure and development challenges of scaling AI from single-threaded loops to distributed, multi-agent workflows capable of parallel human-level judgment. Participants are tasked with solving critical pain points including high-throughput reliability, cost control, prompt engineering for parent-subagent hierarchies, and the security of untrusted context inputs. The ultimate objective is to establish the tooling necessary to make operating fleets of autonomous agents as routine and reliable as deploying standard web services or executing Spark jobs.

The First 10-person, $100B Company

Y Combinator is actively funding small, high-agency teams to replicate the capital efficiency of the early cloud computing era, leveraging new AI tools to build multi-billion dollar companies with as little as $500,000. The fund explicitly targets creating the first 10-person startup valued at $100 billion by prioritizing revenue per employee and exploiting the speed advantages smaller groups have over bloated incumbents. This strategic shift aims to redefine startup growth by proving that ambitious founders can scale rapidly with minimal personnel and overhead.

The Finance Startup Bringing Agentic AI to Wall Street

Arnie Englander, Gustaf Alstromer, Chas Englander

Model ML, founded by siblings Arne and Chas, has transitioned its financial services agentic AI platform from a testing phase to rapid deployment, securing contracts with approximately 10% of the world's largest investment institutions. By replicating human cognitive access within familiar interfaces like Excel, the system automates complex data extraction and document creation for C-suite executives, shifting purchasing cycles from decade-long software licenses to annual multi-year agreements. With a global engineering presence and a strategy prioritizing high-trust in-person demos, the founders leverage their previous exits from Fat Llama and Fancy to drive a market where autonomous tasks increasingly replace traditional user interfaces.

Video Generation as a Primitive

This event addresses the rapid evolution of video generation from high-quality clips to a near-zero cost, ubiquitous computing primitive that enables the on-the-fly creation of any visual content. It outlines transformative applications across media, commerce, gaming, and robotics, while highlighting Y Combinator's call for founders to build infrastructure for this new landscape rather than just content outputs. The discussion emphasizes a fundamental industry shift toward infinite, low-latency video tools that allow for personalized experiences and novel interactions that were previously impossible.

Retraining Workers for the AI Economy

The US government's new AI Action Plan addresses the critical shortage of skilled tradespeople needed for AI infrastructure by directing the Departments of Labour and Commerce to fund rapid retraining programs and startups that develop multimodal AI tutors. These technologies, including voice coaching and AR/VR simulations, aim to compress vocational training timelines from years to months while overcoming the scalability limits of traditional human instruction. By enabling infinite, affordable training at this specific intersection of physical labor and artificial intelligence, the initiative invites startups to build solutions that prepare workers for high-demand roles in semiconductor and data center construction.

Scaling and the Road to Human-Level AI | Anthropic Co-founder Jared Kaplan

Jared Kaplan, Diana

This event details how precise scaling laws governing pre-training and reinforcement learning enable predictable AI progress, with task horizons doubling every seven months to support long-term autonomous work. It highlights the launch of Claude 4, which introduces persistent memory and refined coding agents to facilitate complex, multi-day tasks while advocating for a strategic focus on "70% right" applications and greenfield domains like law and finance. The discussion concludes with a roadmap for shifting human-AI collaboration from supervisory oversight to full automation, driven by the expectation that incremental self-correction and hybrid training data will exponentially extend AI capabilities.

Brand Design Tips From Linear Founder Karri Saarinen

Karri Saarinen, Aaron Epstein

Kari Saarinen outlines how early-stage startups should align their brand authenticity with their development stage, advising against mimicking mature companies like Stripe or Linear before product readiness. She critiques five specific landing pages for issues such as distracting visuals, vague messaging, and poor audience segmentation, recommending targeted copy and clearer calls to action to improve conversion. The analysis emphasizes that website design must evolve from abstract mystery for waitlists to detailed utility for enterprise sales, ensuring every visual element serves a functional purpose rather than adding noise.

How Replit Went From $10M to $100M ARR In Just 9 Months

Amjad Masad, Tom, Dave

Replit has strategically pivoted to an AI-driven mission centered on its autonomous Replit Agent, a move supported by a dedicated infrastructure of NixOS and parallel agent sampling that achieved 45% monthly growth following a targeted workforce reduction. CEO Amjad Massad describes this transition as lowering barriers to software creation, enabling product managers and designers to deploy applications directly and shifting the primary bottleneck from engineering capacity to the volume of human ideas. As the technology matures beyond model coherence into complex computer use, the company anticipates a future where generalist knowledge and idea generation replace traditional coding syntax as the critical skills for creators.