Latest Interviews

Self-Improving Harnesses, Local Personal AI And YC's Agent For Work | YC Paper Club

Seth Karten, Jon Saad-Falcon, Josh France, Regan Bell

YC Harness Club researchers demonstrate that "self-improving harnesses" capable of modifying their own code and memory have driven Arc AGI success rates from 30% to 100%, surpassing standard prompt engineering by leveraging recursive learning architectures. Presentations by Seth and John Sad Falcone detail specific implementations like Prime Agent's multi-tier context management and OpenJarvis's local on-device stack, which collectively achieve frontier-level performance while reducing costs by 800x and enabling autonomous long-horizon research. Furthermore, the QM project establishes a centralized cloud-hosted framework with dynamic sandboxing and fine-grained permissions, addressing critical challenges in system-wide coordination and data security for scalable multi-agent deployment.

Max Junestrand: You Need The Willingness To Learn Faster Than Anyone Else

Max Junestrand, Gustav Omstrammer

Legora has scaled from three Swedish engineers to over 750 employees, generating $100 million in annual recurring revenue while deploying an agentic AI operating system that now serves three percent of the world's lawyers. Following a strategic pivot after a $35 million funding round revealed revenue lagged behind interest costs, the company shifted focus from growth velocity to product-market fit by freezing sales and transitioning to proactive, autonomous agents. Founded by Max, August, and Sigge, the firm differentiates itself through a high-intensity culture designed to overcome Scandinavian humility, recently securing talent at a 23-year-old salesperson who generated over $10 million without prior industry experience.

Max Hodak: Average Is Not Good Enough

Max Hodak

Max Hodak, CEO of deep-tech startup Science, details the company's solar-powered retinal prosthesis that successfully enabled a patient to read a 300-page novel following major clinical trials. The discussion outlines Science's operational strategy, which includes the proprietary Helix purchasing platform to accelerate hardware procurement and an Eigenreviews performance system utilizing eigenvector centrality for continuous employee evaluation. Hodak further argues that success in brain-computer interfaces relies on high-speed iteration, mission-driven leadership, and building custom internal tools rather than relying on generic enterprise software.

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.

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.

World Models, JEPA And The Path To Sample-Efficient RL

Ankit, Francois

This event analyzes the critical bottleneck of sample efficiency in artificial intelligence, contrasting current deep learning models' massive data requirements with the human brain's ability to learn from minimal experience through superior world modeling. The discussion details how advancements in non-differentiable control theories, video diffusion architectures, and Joint Embedding Predictive Architectures are shifting strategies from model-free behavior cloning to synthetic, simulation-based planning for complex robotic and autonomous driving tasks. By addressing scaling challenges in high-dimensional action spaces and architectural limitations like the Transformer's inefficiency in time-domain compression, the presentation outlines a roadmap toward general-purpose robotics and AGI by 2026 through the integration of "awake sleep" mechanisms and physics-informed predictive systems.

How The Best Companies Defend Against Mediocrity And Rot

Garry, Eric Ries

Eric Ries exposes how the 1980s doctrine of shareholder primacy systematically undermines long-term company value, citing cases like Jeff Lawson's ouster at Twilio and Saul Price's departure from FedMart as evidence that founder control is often lost to short-term investor demands. To counter this instability, the event advocates for structural reforms such as the industrial foundation model, which legally prioritizes mission over profit, as demonstrated by the sustained success of entities like Novo Nordisk and the mission-preserving safeguards of Anthropic. The presentation concludes by urging founders to immediately adopt Public Benefit Corporation status and design "governance fortresses" to insulate their organizations from the inherent conflicts of standard corporate governance.

The Future Of Brain-Computer Interfaces

Max Hodak, Garry

Science co-founder Max Hodak outlines a biotechnology roadmap centered on the Prima retinal implant, which has already restored form vision to over 40 patients with macular degeneration, and predicts an imminent "takeoff era" that could enable the first 1,000-year lifespans. Beyond vision restoration, his company is developing bio-hybrid neural interfaces that integrate stem-cell-derived neurons into the brain and portable "Vessel" perfusion systems to revolutionize organ preservation and treat severe disabilities like glaucoma. These initiatives aim to shift healthcare from slow drug discovery to precise neural engineering, potentially achieving high-bandwidth human-AI integration within the next two decades.

We're All Addicted To Claude Code

Calvin French-Owen

Industry analysis indicates a decisive shift toward Command Line Interfaces as the primary interaction layer for coding agents, with Anthropic's Cloud Code utilizing subagent models and OpenAI's Codex employing periodic compaction to manage context windows. Senior engineers are increasingly adopting a managerial role to direct autonomous workflows, while top-tier practitioners distinguish themselves by enforcing rigorous testing protocols and minimizing codebase noise to prevent model degradation. This evolution fosters a tension between rapid, bottom-up grassroots adoption driven by speed and enterprise-mandated top-down security controls, fundamentally reshaping software development structures toward smaller teams operating under vast agent armories.

The FDE Playbook for AI Startups with Bob McGrew

Bob McGrew, Jared

Originating at Palantir to solve intelligence agencies' workflow opacity, the Forward Deployed Engineer model embeds analysts and prototypers at customer sites to drive product discovery through high-touch, outcome-based engagements. This strategy has rapidly spread to the AI sector, where over 100 startups now hire FDEs to bridge the gap between generic large-scale models and specific enterprise needs while securing contracts tied to delivered value rather than usage. As OpenAI’s Bob McGrew observes, these on-site teams function as essential execution layers that extract practical utility from rapidly advancing technology, positioning human ingenuity and adaptive implementation as the primary drivers of market adoption.

How This 25-Year-Old Built A $675M Legal AI Startup (With No Legal Experience)

Gustaf Alströmer, Max Junestrand, Mark Mandelmann, Martin Splitt

Founded in 2024 by Max Giunestrand, Legora is an AI-powered workspace for legal professionals that recently secured an $80 million Series B funding round led by Iconic and General Catalyst. The company differentiates itself through a "software + service" hybrid model and a data-centric architecture that ensures GDPR compliance by hosting all European client data within the EU. With a flat organizational structure that scaled from ten to 100 employees in less than a year, Legora is aggressively expanding its operations to hubs in New York, London, and Stockholm to support law firms in navigating the AI transition.

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.

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 To Build The Future: Sam Altman

Sam Altman, Garry Tan

Sam Altman characterizes the current era as the optimal time for launching technology companies, projecting that Artificial Superintelligence could emerge within a decade if deep learning progress continues to compound. He outlines a five-level roadmap where AI evolves from basic chatbots to autonomous organizational managers, emphasizing that energy abundance and rapid iteration are the critical inputs for unlocking material prosperity. While acknowledging past strategic pivots and leadership turbulence at OpenAI, Altman maintains that speed and conviction allow startups to outmaneuver established corporations in achieving AGI.

Building The World's Best Image Diffusion Model

Suhail Doshi, Garry

Playground Labs founder Suhail Doshi launched "Soda," a novel image generation model built from scratch with an 8,000-token context window to enable precise commercial design control. The platform strategically pivots toward professional utility by replacing text-first interfaces with a visual workflow that targets the $2.3 billion Canva market while avoiding the "toy" segmentation of consumer tools. Backed by Doshi's previous SaaS and browser experiences, the hybrid startup combines rapid commercial shipping with dedicated research to refine spatial reasoning and text accuracy for non-designers.