Y Combinator

The Next Breakthrough In AI Agents Is Here

Garry Tan

Chinese startup Manus has officially launched a general-purpose AI agent that leverages Anthropic's Claude 3.7 Sonnet to orchestrate a multi-agent system capable of executing complex real-world tasks. The platform achieved an 86.5% score on the GAIA benchmark and offers transparent, customizable workflows at a cost of approximately $2 per task, effectively positioning itself as a sophisticated alternative to opaque single-model competitors. While critics question its reliance on existing foundational models, co-founder Yiqiao Peek Ji defends the strategy as a deliberate focus on architectural innovation and user experience rather than model development.

Startups that make it are rare. You can do it alone, but with a good co-founder you can go farther.

The discussion emphasizes that successful startups depend on highly aligned co-founders capable of navigating critical challenges, rejecting the notion that working alone is a safer alternative to partnership. This perspective aligns with the Renee Adler philosophy that all business problems are ultimately people problems, requiring founders to actively manage conflict and collaboration to secure necessary capital and talent. Ultimately, founders who refuse to engage in these essential interpersonal dynamics risk ceding market dominance to competitors who are willing to "play to win" by building strong teams and resources.

The idea for DoorDash didn’t come from surveys—it came from living a day in the customer’s shoes.

DoorDash's founders identified a critical delivery gap after observing a macaroon store owner reject a thick booklet of lost orders, revealing a scalable opportunity for the broader restaurant and retail ecosystem. This insight drove their strategy to validate three distinct components: the merchant need, consumer demand, and driver supply, which proved essential during their Y Combinator period. The resulting platform successfully addressed the unmet revenue potential for small merchants while establishing the necessary infrastructure for a functional gig economy workforce.

What Founders Can Do To Improve Their Design Game

Aaron, Raphael Schaad

Raphael Schad argues that successful startups require the intersection of design, business, and technology, urging designers to learn coding to bridge the gap between concept and execution. His startup Cron exemplifies this approach, having evolved into Notion Calendar following an acquisition that validated the power of founders who can build their own product visions. As artificial intelligence reshapes product categories, Schad emphasizes that the ability to create interactive prototypes and understand technical constraints remains a critical competitive advantage for modern founders.

DoorDash CEO: Customer Obsession, Surviving Startup Death & Creating A New Market

Garry, Tony Xu

DoorDash was founded on the principle of empowering physical businesses after observing a macaroon store owner's inability to fulfill delivery requests, leading to a validated strategy that prioritized suburban markets and a diverse driver workforce over traditional high-density city centers. The company weathered critical operational crises, such as the 2013 Stanford delivery delays and persistent investor skepticism, by establishing a customer-obsessed culture that later enabled it to surpass competitors during the 2020 pandemic through strategic commission cuts and industry-wide support. Today, DoorDash employs over seven million drivers and commands approximately 60% of the U.S. food delivery market share while maintaining its long-term mission to grow the GDP of every city.

Perplexity's competitive advantage against Google comes down to user obsession and product taste.

Aravind Srinivas

Future competitive differentiation hinges on an obsession with the user and strong product taste, a combination currently held exclusively by Google among major entities. Despite generating $200 billion in annual search revenue alongside $100 billion from YouTube and Cloud, the company faces structural conflicts where advertising interests often overshadow search utility. While search remains the sole high-margin profit driver, other divisions face distinct financial constraints, with Cloud only recently achieving profitability and YouTube unable to reach high margins due to creator payouts and subscription ad exclusions.

Navigating co-founder relationships is a journey that takes self examination and discovery.

This discussion examines how high-pressure startup environments and immigrant backgrounds often condition co-founders to suppress dissent and prioritize stability, leading to extreme work habits and internal conflict. The speakers identify these intense relational dynamics as catalysts for deep self-examination, revealing that the resulting emotional strain is a normal, albeit difficult, aspect of the entrepreneurial journey. Ultimately, the dialogue frames the ongoing resolution of these internalized patterns not as a final achievement but as a necessary process for healing and eventual growth.

How Top 1% Founders Navigate Co-founder Conflict

Harj, Patrick Collison, Garry, Gary, Jared Harge, Diana

Founders often face severe psychological strain and destructive co-founder dynamics, such as self-abandonment and cultural clashes, which experts argue require shifting from servant leadership to asserting personal authority for effective conflict resolution. To support technical talent, YC is simultaneously launching Summer Fellows Grants that award $20,000 in cash and $90,000 in compute credits to undergraduate students working on novel AI projects, while providing access to AI Startup School and in-person co-working sessions. The discussion concludes that while solo founders avoid interpersonal friction, they are at a significant disadvantage against the superlative collaboration necessary to build exceptional, scalable companies.

Dylan Field describes design as “art as it applies to problem solving.”

Dylan Field

The event establishes design as the intersection of art and problem-solving, highlighting the current failure of AI to effectively bridge diffusion models and large language models in this domain. It identifies the industry's struggle to integrate these technologies as a primary barrier to automated design. Furthermore, the discussion underscores the continued necessity of human designers, whose value lies in synthesizing broad context, economic dynamics, and user intent to solve complex problems.

From A Pivot To Building A $9.6 Billion Payroll Company

Harj Taggar, Joshua Reeves, Hodge Tagger

Founded by three co-founders who pivoted from a failed expert advice marketplace, Gusto revolutionized the fragmented U.S. payroll industry by delivering a cloud-based, design-driven solution that replaced outdated manual processes used by major incumbents like ADP. The company initially targeted California startups with bare-bones tax filings before leveraging a landmark seed round from prominent tech founders to scale into benefits and compliance, ultimately positioning itself as an AI-powered back-office partner for small businesses. Today, Gusto continues to drive market growth by expanding its self-serve platform and launching a comprehensive compliance hub, aiming to increase the survival rate of new employers through proactive, pattern-matched operational guidance.

Figma's Dylan Field: Exploring the idea maze, vibe coding, and the power of “locking in”

Dylan Field, Garry Tan

Co-founded by Dylan Field, Figma evolved from a series of early pivots into a dominant collaborative design platform that successfully integrated engineers and product managers alongside traditional designers. The company's trajectory was shaped by a user-driven roadmap that birthed features like FigJam and an internal culture of "Maker Weeks," which emphasized rapid iteration over long-term planning. Field's current strategy leverages AI strictly as a capability-enhancing tool to lower participation barriers while acknowledging its inability to replicate human judgment in complex design contexts.

GPT-4.5 = Big Model Energy | YC Decoded

Tom Blomfield

OpenAI has released GPT-4.5, a significantly larger and more human-like model that surpasses its predecessor in emotional intelligence and creative fluency while facing higher costs and specific limitations in complex STEM reasoning. Market reception remains measured as benchmarks show incremental rather than revolutionary improvements, leading researchers to predict a future convergence of general and reasoning capabilities in the upcoming GPT-5 architecture. Simultaneously, YC is hosting a free AI Startup School in San Francisco featuring industry leaders such as Elon Musk, Sam Altman, and Satya Nadella to support emerging AI talent.

Vibe Coding Is The Future

Andrej Karpathy, Gary, Jared Harge, Diana, Abhi von Copycat, Mark Mandelmann, Yoav, Leslie Kendricks, Francesc Campoy Flores, Mark Mirchandani, Melanie Warrick, Trevor, Mark Blyth, Anders Ericsson, Malcolm Gladwell, Picasso, Max Levchin, Toby Lutke, Mark Zuckerberg

Founders characterize "vibe coding" as the emerging dominant standard, where exponential velocity gains from AI tools like Cursor shift the primary engineering role from syntax generation to product judgment and taste. While this paradigm democratizes initial prototyping by generating over 95% of code automatically, participants note a critical divergence where scaling to massive user bases still demands classically trained architects to handle complex systems engineering. Consequently, the industry is pivoting toward hiring assessments that prioritize reviewing capabilities and architectural decision-making to distinguish between the "good enough" code produced by AI and the exceptional quality required for world-class systems.

AI Interfaces Of The Future | Design Review

Aaron Epstein, Raphael Schaad

This event analyzes the industry shift from static, noun-based interfaces to dynamic, verb-centric workflows driven by AI agents and voice technologies. Speakers highlighted emerging tools like VAPI, Gumloop, and AnswerGrid, which prioritize low-latency voice interaction, visual workflow canvases, and parallel data processing to enhance system autonomy and trust. The discussion concludes by projecting a decade-long transition toward touch-first design patterns and standardized canvas models that will fundamentally redefine software interaction paradigms.

Gustaf’s Startup Framework For Spending Money

Gustaf

This analysis argues that startups should avoid modeling themselves after large enterprises, as early-stage organizations lack the functional departments necessary for such structures. The singular focus must remain on discovering product-market fit, since premature adoption of roles like dedicated advertising or specialized offices creates operational distractions that hinder core validation. Founders who prioritize mimicking enterprise maturity risk wasting vital capital and energy on activities that do not contribute to establishing a viable business model.