60 Y Combinator Video Summaries, Bullet Points, TL;DWs & More to Copy & Paste - Page 3

How to Make Claude Code Your AI Engineering Team

Garry Tan's GStack turns Claude Code into an AI engineering dream team! 🚀 Validate ideas, automate design, code review, & QA to ship 10x faster. This open-source toolkit is changing how we build software. Watch the demo! #AIEngineering #GStack

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

1. AI just changed coding forever! 🤯 Y Combinator's Garry Tan reveals GStack, an open-source toolkit that transforms Claude Code into a full AI engineering team. Get ready to build faster than ever. 2. GStack brings an 'agent era' approach to software. It includes 'Office Hours' for rigorous idea validation, adversarial review to catch issues, and even AI-driven design mockups before you write a single line of code. 3. Beyond planning, GStack automates the heavy lifting. It handles code generation, performs staff-level bug catching with full code review, and even runs browser-based QA tests. 4. Imagine running 10-15 parallel AI coding sessions, each working on different features or projects. GStack enables this 'Level 7 software factory' to ship 10x faster, tackling everything from new ideas to open-source fixes. 5. The barrier to building has collapsed. What will YOU build? Check out the full video from Y Combinator to see GStack in action and unlock your AI engineering potential! #AITools #SoftwareDevelopment

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

[Video opens with fast-paced coding montage, text overlay: 'AI Just Changed Coding FOREVER!'] **Hook (0-5s):** **Garry Tan (Voiceover):** We are in a completely new era of building software! [Visual: Quick cuts showing GStack interface, AI generating code/designs] **Content (5-12s):** **Garry Tan (Voiceover):** Y Combinator's Garry Tan built GStack—it turns AI into your full engineering team! From 'Office Hours' to validate ideas, to automated design and QA, GStack helps you build and ship 10x faster. [Visual: GStack UI showing 'Office Hours' questions, then design options, then code snippets] **Call to Action (12-15s):** **Garry Tan (Voiceover):** Want to build something incredible? [Visual: Text overlay: 'Learn More: Link in Bio' or 'Watch Full Video!'] **Garry Tan (Voiceover):** Check out the full video and GStack now!

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

The advent of AI is fundamentally reshaping the landscape of software development, ushering in what Y Combinator President & CEO Garry Tan terms the 'agent era.' His open-source toolkit, GStack, offers a compelling framework for leveraging AI as a sophisticated engineering team. GStack’s strategic value lies in its ability to streamline the entire product lifecycle: from rigorous idea validation through 'Office Hours' that mirror YC partner sessions, to automated design, adversarial review, and comprehensive QA. This integrated approach not only drastically reduces development time and cost but also enhances product quality and market fit by pressure-testing concepts early. For leaders and innovators, GStack represents a significant leap towards a 'Level 7 software factory,' enabling parallel development streams and faster iteration, ultimately collapsing the barrier to building and fostering unprecedented speed to market.

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

Mind blown! 🤯 Y Combinator's Garry Tan just dropped GStack, an open-source toolkit that lets Claude Code become your entire AI engineering team! Imagine having specialists for everything: idea validation (just like YC Office Hours!), adversarial review to catch flaws, AI-driven design mockups, and even automated QA. It's like having a dream team that helps you build and ship products 10x faster. 🚀 In this video, Garry demos it live, showing how an idea goes from concept to refined plan, complete with design, without writing a line of code. This is truly the 'agent era' of building software. Ready to see the future of coding? Hit play on the full video and let us know what you'd build first! 👇 #GStack #AIEngineering #YCombinator #FutureOfCode #OpenSource #TechInnovation #StartupLife #AItools #ClaudeCode #BuildFast

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

AI just changed coding forever! 🤯 Y Combinator's Garry Tan built GStack to turn Claude Code into your AI engineering team. Validate ideas, design, & QA in minutes! 🚀 No more manual coding. Watch the full video to see how! 👇 #AI #Coding #GStack #Tech #Startup

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

We are in a completely new era of building software, the agent era. The barrier to building just collapsed.

— Y Combinator

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

- GStack, an open-source toolkit by Garry Tan, transforms Claude Code into an AI engineering team. - It offers specialized AI skills for idea validation (Office Hours), adversarial review, design, and automated QA. - The system enables parallel development, allowing multiple AI agents to work on different features simultaneously. - GStack significantly accelerates the software development process, helping users ship products up to 10x faster. - It helps validate and refine product ideas rigorously before any code is written, reducing risk.

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

Garry Tan's GStack toolkit transforms Claude Code into an AI engineering team, accelerating software development from idea validation and design to coding and QA, enabling 10x faster shipping.

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

AI just changed coding forever. Here's how.

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

Imagine you want to build a super cool LEGO castle, but it's really complicated. GStack is like having a team of super smart robot friends. One robot helps you brainstorm the best castle ideas, another draws pictures of what it will look like, another helps you put the LEGOs together perfectly, and a final one checks for any broken pieces. This robot team helps you build your awesome castle much faster and better than if you tried to do it all by yourself!

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

Did you know Garry Tan's GStack, an open-source toolkit for AI engineering, gained more GitHub stars than Ruby on Rails in just three weeks after its creation?

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

Problem: Traditional software development is slow, costly, and ideas often fail without proper validation and efficient processes. Solution: GStack leverages AI to create a full engineering team, validating ideas, automating design, coding, and QA, drastically accelerating development and reducing risk.

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

How do you envision AI engineering teams like GStack transforming the startup landscape over the next five years? What kind of project would you build first if you had an AI engineering team at your fingertips? What are the biggest challenges or ethical considerations you foresee with the rise of AI-driven code development?

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

How to Make Claude Code Your AI Engineering Team Y Combinator President & CEO Garry Tan introduces GStack, an open-source toolkit that transforms Claude Code into an AI engineering team. This video explores how GStack streamlines the entire software development process, from idea validation in "Office Hours" to automated design, coding, and QA. Topics Covered: - 00:00:09 Introduction to Garry Tan and the Agent Era - 00:00:59 Introducing GStack and its rapid adoption - 00:01:30 The power of AI coding and Garry's experience - 00:02:22 GStack: The thin harness fat skills approach - 00:02:48 Live demo: Building a tax app with Office Hours - 00:05:18 The critical "What's the strongest evidence?" question - 00:06:28 Expanding the business model with AI insights - 00:07:14 The conversational nature of GStack Office Hours - 00:08:00 GStack Browser automation and secure interactions - 00:09:31 Planning with GStack: Multiple approaches - 00:11:19 Office Hours: Refining a half-baked idea - 00:13:02 Multi-step adversarial review and auto-fixes - 00:13:53 AI-driven design with Design Shotgun - 00:16:21 The full GStack system and various commands - 00:17:39 Achieving Level 7 software factory with parallel work - 00:18:36 Automating QA with SLQA and SL Browse - 00:19:27 Shipping faster with parallel Claude Code sessions - 00:20:20 GStack for managing open-source projects and PRs - 00:21:14 The future of software building with GStack Key Quotes: - "We are in a completely new era of building software, the agent era." - "The barrier to building just collapsed. The only question left is what are you going to build?" Resources Mentioned: - https://github.com/garrytan/gstack - https://www.conductor.build/ - https://www.ycombinator.com/apply - https://www.ycombinator.com/jobs

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

00:00 Introduction 00:00:59 The Rise of GStack 00:02:22 What is GStack? 00:02:48 Live Demo: Office Hours 00:05:18 Validating Your Idea 00:08:00 AI Browser Automation 00:09:31 Crafting a Plan 00:13:02 Adversarial Review & AI Design 00:17:39 Scaling Development with GStack 00:21:14 The Future of Building Software

https://youtube.com/watch?v=wkv2ifxPpF8

How to Make Claude Code Your AI Engineering Team

1. Explore GStack to leverage AI for your software development, turning Claude Code into a specialized engineering team. 2. Utilize the "Office Hours" skill within GStack to rigorously validate your product ideas before writing any code, saving time and resources. 3. Automate design and QA processes with GStack's tools like Design Shotgun and SLQA to accelerate your development cycles. 4. Consider running multiple parallel AI sessions with GStack to tackle different features or projects simultaneously, boosting your output significantly. 5. Embrace the "agent era" by integrating AI tools into your workflow to collapse the barrier to building and ship products 10x faster.

https://youtube.com/watch?v=wkv2ifxPpF8

How Stripe Built Their New Website

Stripe's Head of Design Katie Dill takes Y Combinator behind the scenes of their new website! Learn why they redesigned, the meticulous process, and how AI is shaping the future of web design. Don't miss this deep dive! #WebDesign #Stripe #AIDesign

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

1. Ever wonder what goes into redesigning a world-class website like Stripe's? Y Combinator sits down with Katie Dill, Stripe's Head of Design, for an exclusive look! 🧵 2. The old site, launched in 2020, still looked good. But Stripe's business and product suite evolved, demanding a new narrative to clearly articulate their expansive offerings. 3. Katie reveals key elements like the 'bento box' product display and the unique GDP counter, designed to build trust and show scale without overwhelming users. 4. Discover Stripe's rigorous iteration process, from perfecting animated waves to crafting AI-generated imagery. They fight the 'gravitational pull to mediocrity' to achieve excellence. 5. A must-watch for designers! Y Combinator explores how AI is speeding up prototyping and experimentation, allowing designers to focus on new interaction paradigms and elevate craft. #Stripe #WebDesign #AIDesign

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

**(Hook)** "Stripe's website redesign: Why change what's already great? 🤔" **(Content)** Y Combinator uncovers the secrets with Stripe's Head of Design, Katie Dill! Their 6-year-old site was still visually strong, but their business evolved. Learn how they crafted a new narrative, built trust with a GDP counter, and used AI to push design boundaries. It's about strategic evolution, not just aesthetics! **(CTA)** "Want to see the full design journey? Click the link in bio for the full Y Combinator video!"

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

In a compelling Design Review, Y Combinator features Stripe's Head of Design, Katie Dill, to dissect the strategic rationale and execution of Stripe's recent website redesign. The discussion highlights a critical business imperative: adapting digital presence to reflect an expanded product suite and sophisticated brand identity, even when existing aesthetics remain strong. Dill elaborates on the meticulous, year-long iterative process, emphasizing the "manifesto" role of a website and the intentional use of elements like the "bento box" product architecture and the distinctive GDP counter to convey scale and trust. A significant portion addresses AI's transformative impact on design workflows, accelerating prototyping and exploration while underscoring the enduring need for human craft, taste, and a relentless pursuit of exceptional user experiences over mere efficiency. The insights offer valuable lessons for organizations navigating brand evolution and technological integration in design.

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

Dive deep into the art and science of web design with Y Combinator and Stripe's Head of Design, Katie Dill! 🎨✨ Ever wondered why a successful company like Stripe would redesign a perfectly good website? It's all about evolution! This video pulls back the curtain on their incredible journey to create a site that truly reflects their expanded business and sophisticated brand. From perfecting animated waves to the innovative "bento box" product display and that mind-blowing GDP counter, every detail is intentional. Plus, see how AI is empowering designers to push boundaries faster than ever before. It's a masterclass in fighting the "gravitational pull to mediocrity"! Which part of the redesign excites you most? Let us know in the comments! 👇 #StripeDesign #WebRedesign #DesignProcess #AIDesign #BehindTheScenes #CreativeTech #UXDesign #YCombinator #KatieDill #WebsiteLaunch

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

Stripe's website got a glow-up! ✨ Y Combinator reveals how Head of Design Katie Dill transformed it. Not just looks, but business needs! Peek at their "bento box" & cool GDP counter. AI helps, but human craft is still 🔑! #Stripe #WebDesign #AIDesign #DesignTok #WebsiteRedesign

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

The gravitational pull is to mediocrity. It is just so easy to accept good enough.

— Y Combinator

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

• Stripe's website redesign was driven by business evolution and an expanded product suite, not just outdated visuals. • The new site aims to be a "manifesto," clearly articulating Stripe's offerings and values through intentional design. • Innovative elements like the "bento box" product display and a live GDP counter build trust and convey scale. • AI tools significantly accelerate prototyping and exploration, allowing designers to focus on refining craft and exploring new interaction paradigms. • Stripe emphasizes fighting the "gravitational pull to mediocrity" and "walking the store" to ensure a high-quality, cohesive user experience.

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

Stripe's website redesign, featured by Y Combinator, was a strategic overhaul reflecting business evolution, with AI tools speeding up the design process while prioritizing human craft.

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

Stripe's website: Why change perfection? The answer will shock you!

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

Imagine your favorite toy box. After a few years, you get lots of new, amazing toys! Your old toy box is still nice, but it doesn't show off all your cool new stuff. So, you get a brand new, super-smart toy box that shows everything clearly and makes it easy to find what you want. That's what Stripe did with their website! They added new features and used smart computer helpers (AI) to make it even better, like getting a super-fast helper to arrange all your toys perfectly.

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

Stripe's old homepage, launched in 2020, was considered so well-designed that it still looked current six years later, proving exceptional design can stand the test of time.

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

Stripe's six-year-old website, while visually appealing, no longer accurately reflected the company's evolved business, expanded product suite, and sophisticated brand identity, leading to a diluted narrative. Stripe undertook a comprehensive redesign, crafting a new narrative with interactive "bento box" product displays, a unique GDP counter, and leveraging AI tools to accelerate iterations while maintaining a high standard of craft and intentionality.

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

• How do you balance maintaining a strong brand identity with the need to evolve your digital presence as your business grows? • What are some creative ways your company is fighting the "gravitational pull to mediocrity" in its design and product development? • With AI tools rapidly advancing, how do you see the role of human designers continuing to adapt and specialize in the future?

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

How Stripe Built Their New Website Y Combinator's Design Review hosts Stripe's Head of Design, Katie Dill, for an exclusive look into the strategic redesign of Stripe's website. Discover the motivations behind updating a visually sound site, the meticulous design process, and how AI is revolutionizing their team's approach to creating world-class digital experiences. Topics Covered: * 00:00 - Introduction to the Stripe website redesign * 00:30 - Reviewing the old Stripe site (from 2020) * 01:30 - Why Stripe redesigned a site that still looked good * 03:00 - The biggest narrative changes needed for the new site * 00:05:22 - "What Is The Point Of A Website?" - Stripe's manifesto * 06:44 - Breakdown of the new landing page and its elements * 07:10 - The unique GDP counter as a trust-building design element * 08:45 - Inside Stripe’s “Bento Box” Approach to Product Storytelling * 10:37 - Why Stripe used modals instead of new pages for progressive disclosure * 11:37 - Adding lively animations that don’t feel distracting * 13:50 - The value of beauty and intentionality in web design * 16:24 - Utilizing AI for brand imagery and speeding up the design process * 17:30 - How a website is like a song: Punctuation and flow * 19:11 - Designing the perfect Stripe wave background * 24:48 - Exploring countless homepage concepts before the final design * 27:10 - How AI is changing Stripe’s design process and role of designers * 35:00 - Fighting the “Gravitational Pull to Mediocrity” in design * 40:00 - "Walking the Store": Stripe’s secret to better product design Key Quotes: * "The gravitational pull is to mediocrity. Like, it is just so easy to accept good enough." * "What is the point of a website anyways? I think one part of it is it's your manifesto whether you explicitly call it that or not." Resources Mentioned:

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

00:00 Introduction to Stripe's Website Redesign 00:30 The Old Site and Why it Needed a Change 03:00 Crafting a New Narrative for Stripe's Evolving Business 05:30 The Purpose of a Website: Stripe's Manifesto 06:44 Exploring the New Landing Page Design 07:10 The GDP Counter: Building Trust Through Unique Data 08:45 Stripe's "Bento Box" Approach to Product Storytelling 11:37 Intentional Animations and the Value of Beauty 16:24 Leveraging AI for Brand Imagery and Design Efficiency 19:11 The Iterative Process of Designing the Perfect Wave 27:10 How AI is Transforming Stripe's Design Workflow 35:00 Fighting the "Gravitational Pull to Mediocrity" 40:00 "Walking the Store": Stripe's Approach to User Experience

https://youtube.com/watch?v=ypzNhwpmOD4

How Stripe Built Their New Website

1. Regularly assess if your digital presence truly reflects your current business and brand narrative, even if visuals still look good. 2. Embrace an iterative design process, allowing ample time for exploration and refinement to achieve truly exceptional results, not just "good enough." 3. Leverage AI tools to accelerate prototyping and generate initial concepts, freeing up human designers to focus on high-level craft, intentionality, and exploring new interaction paradigms. 4. Actively "walk the store" – use your own products and experience your website as a user would – to identify pain points and ensure a cohesive, high-quality experience across all touchpoints. 5. Cultivate a culture that actively fights against "the gravitational pull to mediocrity," consistently pushing for excellence and thoughtful design in every detail.

https://youtube.com/watch?v=ypzNhwpmOD4

Robots Are Finally Starting to Work

Robots are finally ready for their 'GPT-1 moment'! 🤖 Y Combinator hosts Physical Intelligence's Quan Vuong who reveals how cross-embodiment models & cloud control are making robots adaptable & scalable. The future of robotics is here! #Robotics #AI

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

1. Is robotics finally having its 'GPT-1 moment'? Y Combinator's latest episode with Physical Intelligence co-founder Quan Vuong dives deep into why the answer is YES. Get ready for a revolution! 🚀 2. Physical Intelligence is building a foundation model to control *any* robot for *any* task. This 'cross-embodiment' approach allows models to learn from diverse robot data, leading to incredible zero-shot capabilities. 3. Forget on-device compute! PI's robots are cloud-controlled, querying models hosted in data centers. This technical insight dramatically simplifies hardware requirements and accelerates deployment. 4. The cost of starting a robotics business is plummeting. Y Combinator shares how this shift will spark a 'Cambrian explosion' of vertical robotics companies, focusing on specific industry problems. 5. From folding laundry to warehouse logistics, robots are tackling complex real-world tasks. This video is a must-watch for founders & anyone curious about the future of physical AI. #RoboticsAI #FutureTech

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

Hook: "Robots folding laundry? Handling complex warehouse tasks? It's happening NOW!" Content: "Y Combinator features Physical Intelligence, the company creating the 'GPT-1 moment' for robotics. They're building one model to control *any* robot, making them incredibly versatile. With cloud-controlled systems, robotics is becoming more accessible and scalable than ever before. Forget expensive hardware – it's all about data and smart models!" CTA: "Want to see the future? Watch the full Y Combinator video to learn how you can be part of the robotics revolution!"

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

Y Combinator's recent discussion with Physical Intelligence co-founder Quan Vuong illuminates a pivotal shift in the robotics landscape, akin to the 'GPT-1 moment' for AI. Physical Intelligence's cross-embodiment foundation model is enabling robots to perform diverse tasks with unprecedented efficiency, including zero-shot capabilities. A key strategic innovation is the adoption of cloud-controlled robotics, significantly reducing on-device compute requirements and hardware dependency. This approach transforms robotics from a complex engineering challenge into a data and operations problem, drastically lowering the barrier to entry for new companies. The video posits an impending 'Cambrian explosion' of vertical robotics startups, poised to capitalize on these advancements by focusing on specific industry pain points with more affordable and adaptable robot solutions, signaling a profound economic opportunity.

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

Mind blown! 🤯 Y Combinator's latest video with Physical Intelligence's Quan Vuong shows us that robots are *finally* starting to work in incredible ways! From flawlessly folding laundry to handling complex warehouse logistics, these aren't just demos – they're real-world applications. ✨ Physical Intelligence is building a universal model that can control ANY robot for ANY task, ushering in the 'GPT-1 moment' for robotics. Their secret? Cross-embodiment training and cloud-controlled systems that make robots smarter and more adaptable than ever. This means the upfront cost for robotics businesses is dropping, opening the door for a massive 'Cambrian explosion' of new startups! 🚀 If you've ever dreamt of building a robot company, or just want to see the future unfold, this is a must-watch. The future of physical intelligence is here, and it's more accessible than you think! #Robotics #AI #FutureTech #Innovation #YCombinator #PhysicalIntelligence #TechTrends #StartupLife #Automation #GPT1Moment

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

Robots folding laundry?! 🤯 Y Combinator just showed us the 'GPT-1 moment' for robotics! Physical Intelligence is making robots smart enough to do ANYTHING. 🤖 No more crazy expensive hardware, it's all in the cloud! Get ready for a robot revolution! #Robotics #AI #FutureIsNow #TechTok #Automation #MindBlown

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

Our mission is to build a model that can control any robot to do any task that is physically capable of.

— Y Combinator

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

• Physical Intelligence is creating a foundation model for robotics, aiming for a 'GPT-1 moment' that controls any robot for any task. • Cross-embodiment training allows models to learn from diverse robot data, leading to superior performance and zero-shot capabilities. • Cloud-controlled robots are a major unlock, reducing on-device compute needs and simplifying hardware requirements. • The upfront cost of starting a robotics company is dramatically decreasing, enabling a 'Cambrian explosion' of vertical startups. • Real-world applications like laundry folding and warehouse logistics are already demonstrating high levels of autonomy.

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

Robotics is experiencing its 'GPT-1 moment' thanks to Physical Intelligence's universal foundation model, enabling versatile, cloud-controlled robots and making startup creation much more accessible.

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

Robots are finally working. Here's how.

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

Imagine robots used to be like really complicated toy cars that only worked on one specific track. If you wanted them to do something new, you had to rebuild the whole car and track! Now, a company called Physical Intelligence is like making a super smart brain that can control *any* toy car on *any* track, even if it's never seen that car or track before. This makes robots much easier and cheaper to use for lots of different jobs, like tidying your room or packing boxes!

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

Generalist robotic models trained on data from multiple robot platforms can perform 50% better than specialist models optimized for a single robot.

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

Problem: Historically, starting a robotics company was prohibitively expensive and complex due to vertical integration, high hardware costs, and difficult data collection. Solution: Physical Intelligence's foundation model, cross-embodiment training, and cloud-controlled approach drastically lower the upfront cost and technical barrier, enabling a 'Cambrian explosion' of accessible vertical robotics startups.

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

What everyday task do you think robots will master next? How do you envision a 'Cambrian explosion' of robotics companies changing our economy?

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

Robots Are Finally Starting to Work - Y Combinator feat. Physical Intelligence In this episode of The Lightcone, Y Combinator hosts Quan Vuong, co-founder of Physical Intelligence, to discuss how their foundation model is ushering in the 'GPT-1 moment' for robotics. Learn about cross-embodiment training, cloud-controlled robots, and the emerging landscape for vertical robotics startups. Topics Covered: • 00:00 The new robotics startup equation • 00:41 Intro: GPT-1 moment for robotics • 03:05 How AI unlocked robotics (RT-2, PaLM-E) • 06:17 Breakthrough: multi-robot scaling (Open-X) • 09:12 The real bottleneck: data • 13:10 Emergence: zero-shot robot skills • 16:01 Real-world demos: laundry & warehouses • 22:21 Robotics becomes a data + ops problem • 23:16 Cloud-controlled robots (big unlock) • 29:03 How to start a robotics company today • 32:33 The coming explosion of robotics startups • 43:53 What’s still missing (and what comes next) Key Quotes: • "Our mission is to build a model that can control any robot to do any task that is physically capable of." • "The equation I think for starting a robotic business has changed and will continue to change at an accelerating pace because the upfront cost is not that high anymore." Resources Mentioned: [Placeholder for resources mentioned in video]

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

00:00 Introduction to the Robotics Revolution 00:41 Physical Intelligence: The GPT-1 Moment 03:05 How AI Unlocked Robotics (RT-2, PaLM-E) 06:17 Breakthrough: Multi-Robot Scaling (Open-X) 09:12 The Real Bottleneck: Data 13:10 Emergence of Zero-Shot Robot Skills 16:01 Real-World Demos: Laundry & Warehouses 23:16 Cloud-Controlled Robots: A Big Unlock 29:03 How to Start a Robotics Company Today 32:33 The Coming Explosion of Robotics Startups 43:53 What's Still Missing and What Comes Next

https://youtube.com/watch?v=4EsUaur0nsQ

Robots Are Finally Starting to Work

1. Understand the new robotics equation: Focus on identifying specific workflow opportunities and collecting data, rather than building expensive proprietary hardware and autonomy stacks. 2. Embrace scrappy hardware and data collection: You don't need top-tier, expensive robots to start; modern models can compensate for hardware inaccuracies. 3. Leverage mixed autonomy: Start with systems where human intervention is acceptable, allowing for gradual improvement and economic break-even before full autonomy. 4. Explore cloud-controlled robotics: Consider offloading heavy computation to the cloud to reduce robot BOM cost and future-proof hardware choices. 5. Seek collaboration and community resources: The robotics community, including Y Combinator and open-source initiatives like PI 0 and PIO5, are actively working to accelerate progress and enable new ventures.

https://youtube.com/watch?v=4EsUaur0nsQ

BillionToOne Is Solving One of Biotech’s Hardest Problems

From a shared lab bench to a $4B valuation! 🤯 Y Combinator uncovers how BillionToOne cracked the code on prenatal genetic testing and is now chasing the 'holy grail' of early cancer detection. This is biotech innovation at its finest. #Biotech #CancerDetection #YCombinator

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

1. How did a startup go from a tiny lab bench to a $4B company solving one of biotech's hardest problems? Y Combinator shares the incredible story of BillionToOne. 2. The challenge: Detecting a single different base pair out of billions in a blood sample – essentially finding a 'needle in a haystack' amidst immense noise. Traditional methods amplify everything. 3. BillionToOne's genius: They add synthetic DNA markers before amplification. This clever trick lets them mathematically subtract noise, revealing the hidden fetal or tumor DNA signals. 4. Impact: Their prenatal genetic test is now one of the most widely used, and they've launched an early version of a cancer test, with sights set on population-wide early cancer detection. 5. It's a masterclass in interdisciplinary innovation, strategic scaling, and tackling the impossible step-by-step. Watch the full Y Combinator video to dive deeper!

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

Hook: Could a simple blood test really detect cancer BEFORE it spreads? 🤯 Content: BillionToOne, featured on Y Combinator, started by detecting tiny bits of baby DNA in a mother's blood – a test now used by 1 in 11 babies! Their secret? A genius trick with synthetic DNA that cuts through noise, making the impossible possible. Now, they're using that SAME tech to find early signs of cancer. Imagine catching cancer before it even reaches stage one! CTA: This isn't science fiction, it's happening! Watch the full Y Combinator video to see how they're making medical history!

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

Y Combinator's latest feature highlights BillionToOne's remarkable journey from a nascent startup to a $4B valuation, disrupting molecular diagnostics. The company's strategic innovation centered on solving the 'needle in a haystack' problem in genetic testing by utilizing proprietary synthetic DNA for noise reduction during amplification. This breakthrough enabled the commercialization of a highly accurate prenatal genetic test, achieving significant market penetration. BillionToOne is now strategically leveraging this core technology to pursue early-stage cancer detection, a long-sought 'holy grail' in medicine. Their phased approach, starting with a less capital-intensive product to fund more ambitious endeavors, exemplifies effective resource management and market expansion in hard tech.

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

Ever wonder how a tiny startup can make a massive impact? 🧬 Y Combinator takes us inside BillionToOne, a company that started with just half a lab bench and a radical idea! They figured out how to find a single piece of baby DNA among billions in a mother's blood, making prenatal testing safer and more accessible. 🤯 And guess what? They're now using that SAME magic to detect cancer super early from a blood sample! Talk about changing the game! Watch the full video to see how they went from an idea to a $4B company. #BiotechBreakthrough #GeneticTesting #CancerResearch #Innovation #StartupLife #YCombinator #ScienceFacts #FutureOfMedicine #DNA

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

🤯 This biotech company is changing EVERYTHING! From finding baby DNA in blood to catching cancer super early. 🧬 The science is wild! Watch Y Combinator's video to see how they did it! #Biotech #CancerDetection #ScienceTok #Innovation #MindBlown

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

Technically we would have solved the, you know, holy grail of cancer detection.

— Y Combinator

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

• BillionToOne developed a prenatal genetic test by detecting rare fetal DNA fragments in maternal blood. • Their breakthrough involves adding synthetic DNA to samples to mathematically remove noise during amplification. • The company grew from two PhD students to a $4B valuation, processing over 600,000 tests annually. • They are expanding their technology to detect early-stage cancers via liquid biopsies, a major medical challenge. • BillionToOne's success is attributed to an interdisciplinary approach and a strategic, phased product development plan.

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

BillionToOne, featured by Y Combinator, developed a leading prenatal blood test by solving complex DNA detection and is now applying this technology to achieve early-stage cancer detection.

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

They found a needle in a haystack—in your blood. What's next?

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

Imagine your blood is like a huge swimming pool with tiny, tiny clues floating around, like little pieces of paper. Some papers are from a baby, and some might be from a very tiny cancer. It's super hard to find just those special papers! BillionToOne found a clever way to put their own special colored papers into the pool first. Then, when they scoop up everything, they can easily tell which papers are theirs and which are the important clues from the baby or cancer, even if they're super rare!

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

Did you know BillionToOne's prenatal genetic test, a technology that didn't exist a decade ago, now screens 1 in 11 babies born in America?

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

Problem: Detecting rare fetal or tumor DNA in a mother's blood is like finding one specific base pair out of billions, with traditional methods amplifying too much noise. Solution: BillionToOne adds synthetic DNA to samples before amplification, allowing them to mathematically subtract noise and accurately identify target DNA fragments, making previously impossible diagnostics possible.

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

How do you think early cancer detection could revolutionize healthcare as we know it? What other 'holy grail' medical problems do you think could be solved with similar innovative approaches?

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

BillionToOne Is Solving One of Biotech’s Hardest Problems Y Combinator sits down with BillionToOne founders David Tsao and Oguzhan Atay to explore how their biotech startup, valued at $4B, developed a widely used prenatal test and is now aiming for early-stage cancer detection from a blood sample. Topics Covered: - 00:00 Finding One Mutation in Billions - 01:00 Detecting Disease From a Blood Sample - 02:10 From Lab Idea to $4B Company - 03:05 The Breakthrough That Made It Possible - 05:15 Building the First Test From Scratch - 07:30 One Customer and a Sales Crisis - 08:55 Inside the Lab: How It Actually Works - 12:10 Detecting Cancer From Blood - 13:50 The Path to Early Cancer Detection - 16:05 Why Early Cancer Detection Is So Hard Key Quotes: - "Once we are there, I think technically we would have solved the, you know, holy grail of cancer detection." - "This converts a difficult biology problem to almost a simple mathematical problem." Resources Mentioned: https://www.billiontoone.com

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

00:00 Finding One Mutation in Billions 01:00 Detecting Disease From a Blood Sample 02:10 From Lab Idea to $4B Company 03:05 The Breakthrough That Made It Possible 05:15 Building the First Test From Scratch 07:30 One Customer and a Sales Crisis 08:55 Inside the Lab: How It Actually Works 12:10 Detecting Cancer From Blood 13:50 The Path to Early Cancer Detection 16:05 Why Early Cancer Detection Is So Hard

https://youtube.com/watch?v=kkv5rZhrLkc

BillionToOne Is Solving One of Biotech’s Hardest Problems

1. Focus on solving a specific, high-impact problem first, even if it's a stepping stone to a larger goal. 2. An interdisciplinary approach, combining different fields like chemistry and data science, can unlock solutions previously thought impossible. 3. Strategic, phased commercialization (e.g., prenatal to late-stage to early-stage cancer) can provide resources and build credibility for more ambitious projects. 4. Don't underestimate the challenge of sales and marketing; innovative solutions require effective communication to reach patients and physicians. 5. Embrace constraints; being resource-limited can foster innovative, step-by-step strategies for achieving big goals.

https://youtube.com/watch?v=kkv5rZhrLkc

This Startup Catches Fraud at Scale

Huge news! 🚨 @ycombinator spotlights Variance, an AI startup just out of stealth with a $21M Series A! They're building the secret AI agents that automate fraud detection & compliance for Fortune 500s & GoFundMe. Keeping the internet safe, one scam at a time! #AI #FraudDetection #YC

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

1. Ever wonder how platforms like GoFundMe stop scams? @ycombinator introduces Variance, an AI startup just out of stealth with a $21M Series A, building the secret weapon against fraud! 2. Variance spent 3 years in stealth because their work is so sensitive. They're building systems for the 'good guys' that are often exploited by the 'bad guys,' like state-sponsored fraud rings. 3. Their AI agents automate content review, identity verification, and fraud detection at scale. Think verifying GoFundMe fundraisers or delivery drivers, all powered by Variance's tech. 4. A key challenge? Unstructured data. Variance's AI can even scrape old UI dashboards to gather crucial info, making previously manual, slow processes instant and consistent. 5. From preventing physical violence to fighting misinformation, Variance's small 12-person team (with 5 engineers acting like a 25-person team thanks to AI coding agents!) is having a massive impact. Watch the full story!

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

Hook: Ever wonder how platforms like GoFundMe stop scams? (0-3s) Content: Meet Variance, an AI startup just out of stealth with a huge $21M Series A! Featured on Y Combinator, they build secret AI agents that automate fraud detection, content review, and identity verification for major companies. They're the hidden force keeping your online world safe, from verifying fundraisers to catching state-sponsored fraud rings. (3-12s) CTA: Want to know more about this incredible tech? Watch the full video on the Y Combinator channel! (12-15s)

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

In a compelling Founder Firesides episode, Y Combinator features Karine Mellata, co-founder of Variance, discussing their emergence from stealth with a $21M Series A. Variance's core innovation lies in purpose-built AI agents that automate complex risk and compliance functions, including fraud detection and identity verification, for Fortune 500 enterprises and platforms like GoFundMe. The strategic decision to operate in stealth for three years underscores the sensitive nature of their work, where transparency could inadvertently aid adversarial actors. The discussion highlights the critical role of their technology in transforming manual, inconsistent processes into scalable, self-healing systems, effectively combating sophisticated threats like organized fraud groups and misinformation campaigns. This capability not only enhances security but also enables faster product launches and operational efficiency for their clients.

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

Big news from the shadows! 🚀 Y Combinator introduces Variance, an AI startup that just emerged from 3 years of stealth with a massive $21M Series A! 🤯 Ever used GoFundMe or signed up for a gig economy platform? Chances are, Variance's secret AI agents were working behind the scenes to keep you safe! They're automating fraud detection, content review, and identity verification for some of the world's largest companies. Think catching state-sponsored fraud rings, verifying fundraisers, and even preventing physical violence online. It's a game-changer! Their small but mighty team of 12 (with 5 engineers!) is leveraging AI coding agents to achieve output closer to a 25-person team. Talk about impact! 💪 Watch the full Y Combinator interview to dive deep into how they're making the internet a safer place, why they stayed secret, and the incredible journey of building this company. You won't believe what they've overcome! #VarianceAI #YCombinator #AIStartup #FraudDetection #TechInnovation #StealthMode #SeriesA #GoFundMe #CyberSecurity #FutureOfAI

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

This startup was a secret weapon for 3 years! 🤫 Variance just got $21M to fight fraud with AI! They help GoFundMe & huge companies stop scammers. 🤯 Keeping the internet safe, you probably used their tech without knowing! #AI #FraudDetection #SecretTech #StartupLife #YCombinator

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

We're building the systems that are often used by the bad guys, but we're building them for the good guys.

— Y Combinator

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

• Variance, an AI startup, announced a $21 million Series A after three years in stealth. • They build purpose-built AI agents for fraud detection, content review, and identity verification at scale. • Their technology is used by Fortune 500 companies and platforms like GoFundMe to combat sophisticated fraud rings. • Operating in stealth was crucial due to the sensitive nature of their work, preventing bad actors from exploiting their methods. • Variance's small team of 12 leverages AI coding agents, achieving the output of a much larger engineering team.

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

Variance, an AI startup, just exited stealth with a $21M Series A, using AI agents to automate fraud detection and compliance for major platforms and Fortune 500s.

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

This secret AI startup just raised $21M!

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

Imagine you have a magic robot detective that can instantly check if people or stories online are real or fake, just like a super-smart grown-up would. Variance built these robot detectives for big companies so they can stop bad guys from tricking people or doing naughty things online, all without anyone even knowing the robots are there!

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

Variance's AI agents have been instrumental in detecting complex, state-sponsored fraud rings during elections and even preventing potential physical violence by identifying threats online at scale.

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

Problem: Companies struggle to keep up with dynamic, sophisticated fraud and compliance risks using slow, inconsistent manual processes or fragmented deterministic systems. Solution: Variance provides purpose-built AI agents that automate complex content review, fraud detection, and identity verification, creating a self-healing system that adapts rapidly to new threats.

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

What are your thoughts on AI companies operating in stealth for years due to the sensitive nature of their work? How do you think AI's ability to combat misinformation and fraud at scale will change our online interactions in the next decade?

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

This Startup Catches Fraud at Scale In this episode of Founder Firesides, Y Combinator Managing Partner Jared Friedman talks to Karine Mellata, co-founder of Variance (W23), who is coming out of stealth and announcing their $21 million Series A. Variance builds purpose-built AI agents for risk and compliance — automating fraud detection, content review, and identity verification for Fortune 500 companies and platforms like GoFundMe. They discuss why Variance built in the shadows for three years, detecting state-sponsored fraud rings, and the accident that nearly ended the company. Topics Covered: 00:00 Variance's $21M Series A and coming out of stealth 00:49 What Variance is and how their AI keeps the internet safe 01:28 Why Variance operated in stealth for three years 02:57 How GoFundMe uses Variance to stop scams and verify fundraisers 03:59 Understanding how scammers exploit big news events 05:50 Variance's role in verifying identities and businesses online 07:44 The three building blocks of Variance's AI agents 09:28 The hardest technical challenge: dealing with unstructured data 12:07 Why Variance's AI agent approach is possible only now 14:22 Catching organized and state-sponsored fraud groups 16:26 How a tiny team of 12 achieves massive output with AI coding agents 20:18 The origin story: How Karine and Michael met at Apple 22:24 Landing the first customer and proving the AI's value 24:57 Karine's personal challenge: recovering from being hit by a truck 29:36 The unwavering commitment to solving one big problem Key Quotes: "We're building the systems that are often used by the bad guys, but we're building them for the good guys." "You don't need human reasoning anymore. So, you have this fully self-healing system which at scale is really transformative." Resources Mentioned: https://www.variance.com/ https://www.ycombinator.com/apply https://www.ycombinator.com/jobs

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

00:00 Introduction: Variance's $21M Series A & Coming Out of Stealth 00:49 What Variance Is: AI Agents for Risk & Compliance 01:28 Why Variance Stayed Secret for 3 Years 02:57 How Variance Helps GoFundMe Stop Scams 03:59 How Scammers Exploit Big News Events 05:50 Verifying IDs and Businesses Online 07:44 The Building Blocks of Variance's AI Agents 09:28 The Hardest Problem: Unstructured Data 12:07 Why This AI Solution Works Now 14:22 Catching Organized & State-Sponsored Fraud 16:26 Small Team, Huge Output with AI Agents 20:18 The Origin Story: Meeting at Apple 22:24 Landing the First Customer 24:57 Overcoming a Near-Catastrophe 29:36 Sticking to One Big Idea

https://youtube.com/watch?v=JF6XIixstmQ

This Startup Catches Fraud at Scale

1. Understand the strategic value of stealth: For highly sensitive technologies, operating in stealth can be a critical strategy to protect your methods and customers from adversarial exploitation. 2. Embrace AI for efficiency: Variance demonstrates how AI agents can dramatically amplify a small team's output, allowing a 12-person company to operate with the capacity of a much larger one. 3. Prioritize problem-driven innovation: Focus on solving a deeply understood problem with a strong sense of duty, rather than merely chasing trending technologies. This conviction can fuel resilience through challenges. 4. Tackle unstructured data directly: Recognize that complex enterprise data is often unstructured and scattered; build solutions that can aggregate and reason over this data, even by scraping legacy UIs. 5. Seek burning pain points for early adoption: When launching an enterprise solution, target customers with urgent, unsolved problems to overcome initial skepticism and accelerate adoption.

https://youtube.com/watch?v=JF6XIixstmQ