Latest Interviews
Showing 1–15 of 131 transcripts.
Clear all filters- Y Combinator14 min
Dot Plots: How to Actually See What Your Users Are Doing
Founders and enterprise product teams can uncover hidden usage patterns and early churn signals by utilizing dot plots, a visualization method originally derived from PayPal's fraud detection systems. This technique replaces opaque aggregate metrics with granular grids that map individual user activity against time, allowing stakeholders to distinguish between active cohorts and vanity behaviors that traditional dashboards mask. By combining these visual insights with cohort retention curves, organizations ranging from early-stage startups to massive platforms like Google Photos can identify specific feature correlations and usage gaps that drive product iteration and contract renewals.
- Y Combinator14 min
How to Get Your First 10 Customers
This guide outlines a tactical framework for founders to secure their first ten customers by prioritizing manual, high-trust interactions over scalable automation. Early acquisition focuses exclusively on personal networks and in-person efforts, utilizing specific channels like micro-events, targeted Reddit DMs, and value-driven advice requests before ever deploying prospecting tools. Once these foundational deals are won through unscalable work, the strategy shifts to utilizing software like Apollo and Clay to replicate the validated pitch for subsequent growth.
- Y Combinator12 min
Pick One Idea and Go Deep
Founders must abandon the search for a theoretically perfect idea and instead commit fully to a single concept to generate reliable market data and achieve deep expertise. This approach demands a radical identity shift and a "burn the boats" mentality, ensuring leaders can run a customer's business with precision while targeting high-stakes sectors like regulated industries or hard tech. By operating at the frontier of AI capabilities and owning specific outcomes rather than just selling software, entrepreneurs validate their viability through real-world execution or acquire unambiguous failure data necessary for strategic pivots.
- Y Combinator17 min
Conductor CEO Charlie Holtz Walks Us Through His AI Coding Setup
Charlie, co-founder of Conductor, leads a team developing a macOS application that orchestrates multiple coding agents to generate software primarily in TypeScript and Rust. The platform enforces a strict non-terminal workflow where AI handles code generation while humans intervene only through specific review modes to maintain quality and prevent error cycles. With a strategic focus on high-efficiency model selection and a future roadmap emphasizing malleable, modifiable agents, Conductor positions itself as a tool that shifts software value from code itself to human-authored prompts and descriptions.
- Y Combinator13 min
How to Build a Self-Improving Company with AI
The event argues for replacing traditional hierarchical management with recursive AI architectures where organizations function as self-improving loops composed of sensor, policy, and learning layers. It demonstrates this shift through a YC case study where autonomous agents analyzed failures and updated code overnight, enabling companies to transition from headcount-constrained models to token-limited systems. By mandating universal data recording and treating internal software as ephemeral, the presentation defines humans as peripheral "edge interfaces" while AI assumes the role of the central "company brain" for decision-making and execution.
- Y Combinator10 min
How To Build A Company With AI From The Ground Up
This presentation argues that AI must function as a company's central operating system rather than a peripheral tool, driving a shift toward closed-loop systems that render the entire organization queryable for automated decision-making. In this model, traditional middle-management hierarchies are eliminated and replaced by AI agents that execute engineering tasks and manage workflows, allowing human leaders to focus on strategy as founders, builders, or directly responsible individuals. By leveraging these "AI software factories," startups gain a decisive advantage over incumbents to achieve tenfold velocity increases and operate with human headcount replaced by optimized token usage.
- Y Combinator20 min
The Powerful Alternative To Fine-Tuning
Ian Fischer, Mark Mandelbaum, Francesc Campoy Flores, Mark Mandelmann, Jr.
Poetic, led by co-CEO Ian Fisher, deploys a recursive Meta-System that autonomously optimizes foundation models to outperform them on benchmarks like Arc AGI v2 and the Humanity's Last Exam at a fraction of the cost. This stealth-stage platform functions as a cost-effective "stilts" for startups, achieving 9 percentage point gains over competing models while utilizing a lean team of seven scientists rather than massive training budgets. The company now invites early access from ventures seeking to integrate these self-improving reasoning strategies into their specific verticals.
- Y Combinator8 min
The New Way To Build A Startup
Companies such as GigaML, Legion Health, and PhaseShift are redefining operational scale by automating all internal workflows—from code integration to clinical operations—rather than merely augmenting isolated functions. This "20X company" model enables lean teams of just a few employees to service Fortune 500 clients and achieve fourfold revenue growth while maintaining flat headcounts that would traditionally require hundreds of staff. By treating internal processes as the primary product, these organizations achieve AGI-level productivity, allowing them to compete directly with massive incumbents without the corresponding payroll burdens.
- Y Combinator16 min
Inside The Startup Building Reusable Rockets
Aaron Epstein, Andy Lapsa, Tom Feldman
Stokes Space, founded in 2019 by former Blue Origin engineers Andy and Tom, is developing the Nova two-stage rocket to achieve full reusability for both orbital stages and drastically reduce launch costs. Backed by nearly $1 billion in funding and operating from a 168,000-square-foot factory in Kent, Washington, the company plans to execute its first orbital launch later this year from Historic Complex 14 in Florida. The team aims to establish daily launch cadence and ubiquitous space access by integrating in-house manufacturing with a proprietary software platform to manage rigorous testing and FAA-regulated operations.
- Y Combinator12 min
How Intelligent Is AI, Really?
Diana Hu, Greg Kamradt, François Chollet
The ARC Prize Foundation, guided by François Chollet's definition of intelligence as efficient novelty, is advancing its benchmark through the upcoming interactive Arc AGI v3, which removes all textual instructions to test agent generalization via video game-like environments. Major industry players including OpenAI, xAI, Google, and Anthropic have adopted the benchmark as a standard metric, though the Foundation explicitly warns that high scores alone do not constitute AGI and emphasizes efficiency in data and energy consumption over simple accuracy. The organization maintains a rigorous stance that true artificial general intelligence will only be declared when systems demonstrate human-level adaptability and efficiency, immediately analyzing any system achieving perfect benchmark scores for potential overfitting rather than accepting them as a final milestone.
- Y Combinator13 min
Inside The Startup Launching AI Data Centers Into Space
Aaron Epstein, Philip Johnston, Ezra Feilden, Adi Oltean, Addy
Launched on November 2nd via a SpaceX rideshare mission, aerospace startup StarCloud successfully deployed the first orbital data center, StarCloud One, which houses an NVIDIA H100 GPU to operate 100 times more powerfully than previous space computers. Founded in Summer 2024 by CEO Philip Johnston, Microsoft veteran Addy, and NASA alumnus Ezra, the team rapidly pivoted from solar power concepts to orbital compute to leverage cheaper launch costs for massive, sun-synchronous facilities that utilize deep space for cooling. With a second satellite featuring NVIDIA Blackwell architecture scheduled for October 2025, StarCloud aims to scale to 40-megawatt centers that bypass terrestrial constraints on land and water while validating high-performance workloads like model training in the vacuum.
- Y Combinator9 min
Transformers Explained: The Discovery That Changed AI Forever
This event traces the evolution of AI from early neural networks plagued by vanishing gradients to the 2017 introduction of the transformer architecture, which replaced sequential processing with parallel self-attention. Key milestones include the LSTM's ability to model long-range dependencies, Google Translate's adoption of attention-based sequence-to-sequence models, and the subsequent bifurcation of transformers into encoder-focused BERT and decoder-focused GPT series. These developments enabled the shift from single-task specialists to general-purpose large language models, establishing the foundation for current state-of-the-art systems like ChatGPT and Claude.
- Y Combinator8 min
What Everyone Is Getting Wrong About AI And Jobs
This analysis synthesizes historical precedents like containerization and cloud computing to refute extreme predictions of mass unemployment, demonstrating instead that AI efficiency triggers Jevons' Paradox by lowering costs and exploding demand for services. Prominent figures such as Andrej Karpathy and Aaron Levy argue that while AI automates rote tasks, it predominantly refills labor markets by elevating human roles to supervisory positions and addressing pent-up demand in sectors like healthcare and law. Consequently, founders and investors are urged to actively build solutions that leverage this latent demand rather than waiting for policy interventions or succumbing to fatalistic views on economic transformation.
- Y Combinator13 min
This Startup Is Trying To Delete 29% Of All CO2 Emissions
Remora has launched a solid-pellet based carbon capture system that reduces CO2 emissions by up to 90% in hard-to-electrify diesel trucks and freight trains, converting the captured gas into beverage-grade CO2 for commercial reuse. Backed by $117 million in venture capital, the Detroit-based company is vertically integrating its manufacturing to deploy modular retrofits to major partners like Ryder and Union Pacific while testing the technology on a repurposed locomotive. Co-founders Paul Gross and Christina Reynolds aim to scale this point-source solution to capture one billion tons of carbon annually by overcoming the logistical challenges of retrofitting diverse vehicle fleets.
- Y Combinator13 min
OpenAI vs. Deepseek vs. Qwen: Comparing Open Source LLM Architectures
OpenAI, Alibaba Cloud, and DeepSeek have each launched significant open-weight language models featuring distinct Mixture of Experts architectures and advanced long-context capabilities. While OpenAI's GPT-OSS prioritizes inference efficiency on consumer hardware, Alibaba's Qwen 3 introduces flexible dense and sparse variants with dual reasoning modes, and DeepSeek's V3.1 achieves superior memory efficiency through Multi-Head Latent Attention. Despite differing engineering strategies for scaling and alignment, all three families demonstrate comparable performance benchmarks derived from trillions of tokens and complex post-training pipelines.