Y Combinator

How To Design In The Agent Era

Stephen Haney, Aaron Epstein

Led by Stephen Haney, Paper is an AI-native design platform currently ranking as the third most-used tool behind Figma and Sketch, utilizing a native HTML and CSS rendering engine to facilitate precise agent collaboration. The application distinguishes itself by treating code as the single source of truth, offering features like direct React component handoff and an MCP server that allows autonomous agents to manipulate local repositories and design canvases. While the team prioritizes human curation to maintain design quality against common AI aesthetic pitfalls, the platform is rapidly evolving into a visual coordination layer for the broader agent ecosystem by integrating tools like Paper Shaders and multi-model image generation.

Garry Tan: The Future of AGI Is Personal

Garry Tan

The speaker advocates for "Personal AGI," a system of user-owned agents that process vast personal context on private infrastructure to drastically increase output and preserve cognitive independence. Drawing parallels to Baruch Spinoza's rejection of institutional compromise, the talk details how "skill files" and hybrid architectures enable individuals to replace traditional workforces with scalable, compounding digital labor. This shift promises to democratize high-level execution by allowing founders to achieve unprecedented economic leverage while retaining full ownership of their generated knowledge and intellectual tools.

The Case For Data Centers In Space

Philip Johnston, Mark Mandelmann

Founded by Philip Johnston in 2024 and backed by a $170 million Series A from Benchmark, StarCloud has emerged as YC's fastest-growing unicorn by pioneering orbital data centers to bypass terrestrial energy constraints. The company successfully launched its first satellite in November 2025, deploying custom-modified GPU hardware and thermal management systems to run AI models like Gemini while utilizing automotive-grade components to mitigate radiation risks. With plans to scale to 88,000 satellites by 2028 and deploy high-bandwidth compute for government and hyperscale customers, StarCloud aims to eventually generate terawatts of processing capacity through partnerships with SpaceX and NVIDIA.

Waymo Co-CEO Dmitri Dolgov: The Demo Is Only 1% Of The Work

Dmitri Dolgov

Waymo has advanced its fully autonomous fleet to operate 500 weekly trips across 15 U.S. cities, achieving a safety record 17 times better than human drivers through a multimodal sensor architecture and a foundation model utilizing both fast geometric reactions and slow semantic reasoning. The company addresses the unique challenges of physical AI by integrating real-world data into a generative simulation ecosystem that creates rare edge cases for training, while its "Safety and Readiness Framework" rigorously validates performance to ensure public trust and regulatory compliance. Looking ahead, this structural augmentation and flywheel of agent, simulator, and critic data positions Waymo to expand its technology beyond personal vehicles into trucking and broader physical applications.

Patrick Collison: Is AI Breaking the Lean Startup Playbook?

Patrick Collison, Harj Taggar

Patrick Collison shares his journey from dropping out of college twice to launching Stripe, emphasizing the critical role of retaining "cognitive L1 cache" over reliance on AI tools for deep reasoning. He highlights that while new business formation on Stripe has doubled and AI is accelerating revenue milestones, entrepreneurs should remain wary of decentralization fears and instead focus on solving visceral customer problems like legacy payment friction. Collison concludes by advising founders to build fundamental knowledge of first principles and to assess whether their chosen venture is a cause they are willing to pursue for decades.

Kernels and Chips: Cursor, NVIDIA, and Meta Researchers on GPU Performance | YC Paper Club

Stuart Sul, John, Francois Chaubard, Jon Saad-Falcon, Mark Saroufim, Misha Smelyanskiy, Brennan Shacklett

The event synthesizes critical industry shifts toward specialized ASICs and disaggregated architectures that separate training and inference workloads to maximize intelligence per watt. Technical deep dives highlight breakthroughs in multi-GPU kernel optimization and local inference, demonstrating that consumer accelerators can now deliver nearly 90% of frontier model utility while reducing energy costs by 70%. These innovations collectively address systemic bottlenecks in network communication and latency, establishing a roadmap for distributed, cost-efficient AI ecosystems that minimize reliance on centralized cloud resources.

Alexandr Wang: “This is a Once-in-a-Civilization Opportunity”

Alexandr Wang, Garry Tan

Alexander Wang discusses his journey from founding Scale AI to leading Meta's Frontier Lab, where he is developing the Muse Spark model to deliver frontier-level AI capabilities at a fraction of the cost. He argues that the immediate future of the industry will be defined by agentic looping and the orchestration of autonomous agents rather than raw model size, predicting a massive expansion in global GDP driven by these technologies. Wang concludes by advising founders to build independent convictions based on first principles while emphasizing that systematic thinking and vision are becoming the scarce resources in an era of abundant computational intelligence.

Blake Scholl: How 50 People Built a Supersonic Jet

Blake Scholl

Founded by Blake Scholl in 2014, Boom Supersonic successfully developed the XB-1, the first privately built jet to break the sound barrier, using a lean team and software-driven engineering to overcome decades of aviation stagnation. This historic flight directly catalyzed a shift in US federal policy, prompting an Executive Order that legalized supersonic flight over land and clearing the path for the upcoming Overture airliner. By vertically integrating manufacturing and creating revenue through ground-based engines, the company aims to double global travel speeds while eliminating the traditional sonic boom barrier.

Sam Altman: "Never a Better Time to Do a Startup"

Sam Altman, Paul Graham, Garry Tan

Sam Altman and Y Combinator highlight an exponential shift in startup capabilities where coding agents reduce development time to mere seconds while hard tech representation in the portfolio surges to 25%. The organization positions itself as a decentralizing force to prevent single-entity control of AI power, explicitly rejecting the "live action role play" mindset in favor of heretical, high-conviction building. Altman warns of safety risks regarding surveillance and model concentration but maintains that widespread founder agency and distributed innovation will ultimately prevent economic collapse or catastrophic safety incidents.

Boris Cherny: We Cut 80% of Claude Code’s Prompt

Boris Cherny, Diana Hu

The Opus 5 model release demonstrates a 30% score on the ARC AGI benchmark and unprecedented autonomous endurance by eliminating over 80% of legacy system prompts through an empirical ablation methodology. This shift enabled dynamic workflows where thousands of agents autonomously rewrote complex software stacks like the Bunn runtime and Cloud Code desktop app in mere days. Strategic advice for developers emphasizes replacing rigid prompt engineering with context design and verification protocols to leverage these emergent agentic capabilities effectively.

Self-Maintaining APIs

I

A recent analysis of over 50 early-stage vendor relationships and AWS data reveals that undetected API changes cause significant downtime, prompting a shift toward agentic tools that automate codebase access. To solve this infrastructure gap, the proposed solution introduces provider agents that scan customer repositories upon breaking changes and automatically generate pull requests with necessary fixes. The speaker is currently inviting Y Combinator applicants to build these specialized agents or a neutral third-party service that functions like Dependabot but for API migrations.

What Actually Makes A Startup Durable

Paul Graham and YC partners outline a strategic framework for navigating the AI era, emphasizing that while AI increases individual productivity by 1,000x, founders must target "hard" problems with durable moats to survive market saturation. The session advises building companies through co-founders who provide essential emotional support rather than just complementary skills, noting that human judgment remains critical for community building and navigating high-stakes pivots. Finally, students are offered over $25,000 in AI credits to neutralize geographical disadvantages, reinforcing the program's focus on rapid execution and deep market engagement over theoretical development.

AI-Native Compliance Infrastructure

Y Combinator is soliciting founder applications for companies that replace fragmented, manual compliance infrastructure with AI-native systems capable of consolidating regulatory tools and reducing reliance on specialized legal personnel. These ventures aim to solve the inefficiencies of state-by-state regulatory patchwork by automating monitoring, anomaly detection, and audit trails to provide finance teams with real-time global visibility. As such solutions become essential enterprise infrastructure for global expansion, successful applicants will be expected to redefine compliance operations through a default AI operating mode rather than simple process automation.

Proving You’re Human

Following a $25 million fraud involving deepfake avatars impersonating corporate executives, industry experts identify the collapse of traditional human verification as a critical vulnerability across digital communications. Strategic initiatives are now prioritizing the reconstruction of an internet trust layer capable of authenticating identities in video calls, messages, and transactions without compromising user privacy. The entity that successfully builds this scalable verification infrastructure is projected to become the mandatory standard for financial institutions and social platforms over the next decade.

What Big Tech Missed And How Startups Can Still Win

Alexandre LeBrun, Zuckerberg, Jan, Alex

Ami Labs is raising approximately one billion euros in seed funding to develop foundational world models that train on sensory data rather than text, addressing the limitations of current large language models for robotics applications. Co-founded by Alex K. and Yann LeCun following their shared history at Meta and various startups, the Paris-based company aims to build general-purpose robots capable of operating safely in open environments by solving the compute bottlenecks and latency issues inherent in existing vision-language-action architectures. While the venture tackles the high capital demands of securing thousands of GPUs and top-tier talent, it operates under a strict strategic constraint where failure to deliver visible progress within two years could render the company unsustainable due to the intense external expectations created by the massive funding round.