Y Combinator

How To Build The Future: Aravind Srinivas

Aravind Srinivas, David Lieb

Perplexity achieved a valuation exceeding $9 billion by pivoting from enterprise prototypes to a consumer search engine that leverages modern LLMs to summarize top search results while adhering to a "user is never wrong" philosophy. Founder Aravind Srinivas credits the platform's exponential growth to viral user adoption and a strategic focus on engagement metrics rather than complex agent-based browsing. Looking ahead, the company aims to evolve into a full-stack decision engine capable of executing end-to-end tasks like booking travel, thereby overcoming the current industry split between information retrieval and commercial monetization.

How AI Is Changing Enterprise

Aaron Levie, Gary, Jared, Harj, Diana, Mark Mandelmann, Mark Mirchandani, Mark Mandalini, David Eastman, Melanie Warrick

Industry leaders assert that sustainable AI startups must evolve into software companies delivering proprietary business logic rather than relying on simple model wrappers, a shift driven by the economic reality that intelligence is becoming commoditized. This transition is fueled by Jevons Paradox, which predicts that lower costs will expand the total addressable market by enabling enterprises to automate previously unaffordable workflows while shifting pricing models toward usage-based structures. Consequently, Fortune 500 organizations are rapidly adopting AI-native strategies focused on core intellectual property, leveraging agentic workflows to reinvest efficiency gains into growth rather than workforce reduction.

How Blake Scholl Built The First Independent Supersonic Jet

Blake Scholl, Jared Friedman

Boom Supersonic, founded by former Amazon and Groupon engineer Blake Schull, is advancing its XB-1 supersonic prototype to validate the technical feasibility of a next-generation airliner. This initiative aims to replace the Concorde with the Overture, a 65-passenger aircraft utilizing sustainable fuel and sonic boom mitigation technology to enable overland travel. By prioritizing business-class pricing and targeting commercial operations by 2029, the venture seeks to overcome the historical economic failures of supersonic flight through a lean development strategy.

Former OpenAI CRO Bob McGrew thinks that two jobs of the future will be genius and manager.

Bob McGrew

Future AI-integrated industries will likely be defined by two primary roles: the independent "Genius," who leverages tools for high-impact solo innovation, and the "Manager," a CEO overseeing firms comprised mostly of AI agents. While speakers characterize these emerging pathways as optimistic and engaging, they clarify that AI will serve as the core workforce rather than necessitating the complete displacement of human staff. This shift suggests a transformed business landscape where human leaders and creators coexist with autonomous digital employees to drive efficiency and innovation.

The Right (And Wrong) Way To Spend Money At Your Startup

Brad Flora, Pete Koomen, Nicolas Dessaigne, Gustaf Alströmer, Nicola, Gustav

This presentation outlines a strict capital discipline strategy for startups, emphasizing that pre-seed and seed founders must prioritize product-market fit by avoiding premature hiring and restricting spending to absolute essentials. As companies transition to Series A and beyond, the focus shifts to scaling a predictable revenue engine where growth is fueled only by accretive hiring and retention metrics like Net Dollar Retention. Throughout all stages, the guidance warns against mimicking large corporate structures or overspending on branding, urging founders to maintain transparency and use capital strictly as fuel for validated business models rather than a tool to mask fundamental weaknesses.

How To Get AI Startup Ideas

Gary, Jared, Diana, Harj

YC is hosting its inaugural AI Startup School in San Francisco on June 16th and 17th, featuring high-profile speakers like Elon Musk and Sam Altman to guide eligible CS students and recent graduates through cutting-edge venture creation. The program emphasizes deriving viable startup ideas by leveraging unique domain expertise or conducting undercover research in underserved industries, illustrated by case studies of founders who transformed personal observations at companies like Tesla and Cohere into seven-figure ventures. Participants receive full travel support and strategic advice to persist with ambitious AI projects, moving beyond transient trends to build companies with sustainable competitive advantages in rapidly evolving sectors.

The Engineering Unlocks Behind DeepSeek | YC Decoded

Diana Hu

DeepSeek unveiled its open-source R1 reasoning model and V3 base architecture, delivering performance parity with top-tier proprietary systems like OpenAI o1 through novel algorithmic optimizations such as native FP8 training and Group Relative Policy Optimization. The announcement triggered significant market volatility, erasing nearly $600 billion from NVIDIA's capitalization, while demonstrating that software stack efficiency can rival the hardware advantages of major US tech ecosystems despite strict GPU export controls. This breakthrough validates a new trajectory for the industry by proving that high-cost AI development can be drastically reduced, thereby lowering barriers for startups and accelerating the adoption of accessible, locally deployable intelligence.

The Future of Software Engineering

Language models are projected to reduce software development costs to near zero while driving demand for human engineers to shift from direct coding to managing autonomous agent teams. These agents are expected to assume specialized responsibilities spanning quality assurance, operations, and compliance audits, fundamentally transforming the engineering workflow. Consequently, the speaker is seeking funding for startups that provide generalist developers with tools to coordinate these large agent forces for high-volume software delivery.

B2A: Software Where the Customers Will All Be Agents

The entity introduces a "b2a" investment thesis targeting companies that build software for AI agents as primary customers, aiming to shift the internet landscape from automated impersonation to first-class agent citizenship. This strategy prioritizes funding infrastructure, logistics, and legal frameworks that enable agents to autonomously handle payments, travel bookings, and commercial contracts. Mirroring the coexistence of human and algorithmic traders in stock markets, the approach projects a massive expansion of AI-to-AI interactions across the entire internet ecosystem.

AI Coding Agent for Hardware-Optimized Code

NVIDIA's market dominance relies on CUDA's hand-optimized code, but emerging reasoning models like DeepSeq R1 can now generate comparable AI-optimized kernels, offering a strategic pathway to bypass the critical shortage of low-level system programmers. Founders who develop tools to automate this kernel generation can enable broader adoption of competing hardware and reduce ecosystem dependency on NVIDIA. In response to this opportunity, Y Combinator is actively soliciting applications from entrepreneurs building AI-generated kernel solutions.

Vertical AI Agents

Sam Altman

Vertical AI agents, which are specialized software applications built on large language models to automate specific workflows, are emerging as a primary pathway to building billion-dollar startups across dozens of distinct market sectors. Industry experts highlight high-potential use cases ranging from automated tax accountants and medical billers to dedicated phone support agents and quality assurance testers. This strategy has received strong validation from Sam Altman, who explicitly stated he would focus on building vertical AI agents if he were to restart his career at age 24.

Devtools for AI Agents

A venture firm has pivoted its two-year strategy to fund startups disrupting legacy players through next-generation AI agents capable of autonomous decision-making. Leveraging recent reasoning breakthroughs from model releases like "01" and "03," the firm targets investments in developer tools and building blocks that enable the creation of specialized agent teams. This initiative aims to integrate these autonomous systems across industries, empowering users to enhance productivity and creativity while inviting founders to collaborate on shaping the future of software.

Inference AI Infra in the World of Test-Time Compute

As capital allocation shifts from pre-training to inference for advanced models like DeepSeq R1 and OpenAI 01, surging API demand creates a critical infrastructure cost bottleneck for AI applications. Startups addressing this challenge by rebuilding the inference stack through optimized software tooling and efficient GPU workload management are positioned to unlock significant market growth. Y Combinator is currently soliciting applications from founders developing these essential solutions to enable scalable application deployment.

Bob McGrew: AI Agents And The Path To AGI

Bob McGrew, Garry Tan, Gary

Industry leaders identify a shift toward "test-time compute" and reasoning mechanisms as essential to overcoming data bottlenecks and enabling 99% reliable autonomous agents. OpenAI's hybrid research model, which balances leadership direction with individual autonomy and collective credit, serves as a blueprint for scaling intelligence across LLMs and robotics. Future adoption hinges on developing specialized software interfaces and "forward deployed" engineering teams, ultimately bifurcating the workforce into "lone geniuses" leveraging AI and managers running AI-heavy firms.

AI Personal Staff for Everyone

Historically, personalized knowledge work such as legal, financial, and medical assistance remained exclusive to the wealthy due to the reliance on human labor, while software democratized other services like transportation and photo organization. The organization now aims to bridge this persistent gap by developing advanced AI systems capable of automating these specialized roles for the general public. To achieve this mission, the group is actively recruiting collaborators and employees to build "personal AI staff" products that will deliver elite-level services to billions of users.