
41 segments available
Varun Mohan didn't set out to build one of the fastest-growing AI developer tools. He just knew his company had to change, or die. After initially betting on GPU virtualization, he saw the writing on the wall: if there was a future for his company, it would be at the AI application layer, not infra. Over a single weekend, he and his team pivoted to building Windsurf— a tool to help everyone, both technical and non-technical, write code faster and smarter. In this conversation, Varun shares the inside story of how they pivoted at the edge of failure, trained new models from scratch, outpaced giants like GitHub Copilot, and what the future of building with AI looks like. Apply to Y Combinator: https://ycombinator.com/apply Work at a startup: https://workatastartup.com Chapters (Powered by https://chapterme.co/) - 0:00 Intro 00:53 Windsurf - how big is it, where did it start? 3:00 The big pivot 6:20 Irrational optimism + uncompromising realism 7:52 Earliest versions shipped 10:26 The first customers 13:13 The transition from Codeium to Windsurf 19:45 Going up against Github Copilot 23:15 All insights depreciate; you need to keep proving yourself 26:50 Strong evals go a long way 30:15 Windsurf for hardcore engineering 31:55 Tips to get more precise changes when vibe coding 35:15 How will Windsurf evolve 38:00 Will AI become the infinite workhorse? 38:48 How does Windsurf interview candidates? 42:48 What happens if we get “just in time” software? 44:46 How many non-developers use Windsurf? 47:28 Thoughts on the GPT wrapper meme 49:17 Advice for new AI startups 51:39 Outro
Varun Mohan emphasizes the necessity for startups to continuously innovate and adapt. He discusses how insights can quickly become outdated, using Nvidia as an example of a company that must innovate to stay ahead. He introduces the concept of a 'builder' as a broader definition of a developer, highlighting the democratization of software development.
"One of the things that I think is is true for any startup is you have to keep proving yourself. Every single insight that we have is a depreciating insight. You look at a company like Nvidia. If Nvidi..."
Varun shares the origin story of Windsurf, initially founded as Exofunction, a GPU virtualization company. He explains the shift in focus to AI applications as the landscape changed, particularly with the rise of transformer models. This pivot was crucial for the company's survival and growth.
"Welcome back to another episode of the Light Cone. Today we've got a real treat. We have the co-founder and CEO of Windsurf, one of the people who literally brought Vibe Coding into existence, Verun. ..."
In a pivotal moment for Windsurf, Varun recounts how he and his co-founder decided to pivot the company over a single weekend. Faced with the realization that their initial business model was becoming obsolete, they quickly transitioned to developing a new product, Kodium, to meet the evolving needs of developers.
"ago. uh we didn't start as windsurf uh we actually started as a company called exofunction we were a GPU virtualization company at the time uh previously me and my co-founder had worked on autonomous ..."
Varun discusses the importance of team enthusiasm during the pivot to Kodium. He highlights that choosing a direction that excites the team is crucial for success. The decision to focus on developer tools was driven by their collective experience and belief in the potential of AI-assisted coding.
"weekend me and my co-founder had a conversation along the lines of I don't think this is going to work. We don't think we know how to scale this company. Um and uh at the time, we were early adopters ..."
Varun reflects on the market dynamics that led to their pivot, particularly the emergence of generative models like GPT-3. He explains how these models changed the landscape for deep learning applications, making custom models less relevant and prompting Windsurf to adapt quickly.
"the thing that's remarkable is when you started the company, you had this thesis where you were betting that a lot of companies were going to build their own custom deep learning pipelines to train bi..."
Varun elaborates on the risks associated with their pivot decision, including the potential loss of their existing business. He emphasizes the need for conviction and a clear vision when making such significant changes, highlighting the importance of being all-in on their new direction.
"models these generative models handle so many of the use cases right uh maybe an example is in the past you would train a BERT model uh to actually go ahead and do sentiment classification. But very q..."
Varun discusses the dual mindset required for startup success: irrational optimism and uncompromising realism. He explains how these contrasting beliefs can drive innovation while also ensuring adaptability to changing circumstances.
"would fail we'd fail immediately. We came with an opinionated stance of we were early adopters of a product called GitHub copilot. We thought that that was the tip of the iceberg on where the technolo..."
Varun shares the rapid development of Windsurf's first product version after the pivot. He describes how they built a VS Code extension and launched it within two months, despite initial limitations compared to competitors like GitHub Copilot.
"irrational optimism we basically said hey we know how to run and train models ourselves. We actually trained the first autocomplete models ourselves and ran it on our product gave it out for free. I d..."
After launching their initial product, Varun explains how Windsurf transitioned to training their own models. He discusses the challenges and learning curve involved in developing a competitive AI coding assistant, leveraging their existing infrastructure and expertise.
"beginning of 2023, I'd say the autocomplete capabilities were much better than what co-pilot had. Was that a totally new capability for you guys? Cuz like you guys have been building GPU infrastructur..."
Varun describes how Windsurf attracted early users by offering their product for free. He highlights the interest from various companies looking to integrate the tool into their workflows, showcasing the demand for innovative coding solutions.
"because you were running your own models you started getting interesting customers right? Yeah. So basically what happened the product was free at the time. So we ended up getting a lot of developers ..."
This segment covers the transition from Codeium to Windsurf, highlighting the rapid growth in revenue and the challenges of adapting to a fast-paced industry. Varun reflects on the importance of testing hypotheses and the need for continuous innovation in the AI space.
"Why? I think maybe the fundamental reason that that we thought was quite critical is if we were going to work with companies, companies have developers that write in many languages. Uh like for instan..."
Varun discusses the nature of experimentation in the tech industry, emphasizing that not all bets will succeed. He shares insights on the significance of learning from failures and the need to stay at the forefront of technology to remain competitive.
"make this transition. How about the transition from Kodium to WinSurf? At the time we, you know, now now we're probably going to middle of 2023. We start working with some very large enterprises. With..."
In this segment, Varun reflects on the evolution of AI agents and the challenges faced in developing them. He discusses the advancements in technology that allowed for better agent capabilities and the decision to create Windsurf to enhance developer experiences.
"beginning of last year and and they just didn't work. But there were different pieces we were building that we felt were going to be important to making agents work which is understanding large code b..."
Varun shares the timeline of launching Windsurf, detailing how the team managed to ship the product within three months. He highlights the initial reception and the challenges of refining the product based on user feedback.
"you did that was you forked VS code for VS code was got a whole new set of capabilities that you guys had to learn like basically how to like develop on this VS codebase that I'm sure is super complic..."
This segment focuses on the team dynamics at Windsurf during the product development phase. Varun explains how the engineering team, despite being small, worked collaboratively to pivot and deliver the product quickly.
"And what what happened like was did it like take off immediately or was it unnoticed for for a long time? It took off pretty quickly I would say. Um I think the speed at which it took off among early ..."
Varun discusses the unique sales strategy of Windsurf, emphasizing the importance of a strong go-to-market team. He explains how they engage with large enterprises and the need for personalized support to ensure successful adoption of their technology.
"were working on in the past and work on it immediately. And at that time, how how big uh were you guys? The engineering team was probably still less than 25 people. Wow. This is crazy. Interestingly, ..."
In this segment, Varun highlights how Windsurf empowers non-technical users to create applications without extensive coding knowledge. He shares a success story of a non-technical user who effectively utilized Windsurf to streamline processes within the company.
"nontechnical person who leads u uh like partnerships uh at the company. Uh he has actually replaced buying a bunch of sales tools uh inside the company. uh and uh this is one of those things where I t..."
Varun reflects on Windsurf's approach to competition, emphasizing that the company's morale remains unaffected by competitors. He discusses the importance of having a long-term strategy and the flexibility to adapt, even when facing established players like GitHub Copilot and emerging startups like Cursor.
"might be a weird thing about our company, but our company just doesn't have like like morale is not really affected by what other companies do right? Uh it's not possible if uh like our company has go..."
Varun shares Windsurf's innovative approach to coding tools, contrasting their focus on agentic editing with the traditional chat and autocomplete functionalities of competitors. He explains how they prioritize understanding developer intent and making rapid changes to code, setting Windsurf apart in the market.
"the way we've always Do you educate yourself at all on the competitor's products, though? Yeah. Yeah. I think we don't want to put our heads in the sand and kind of tell ourselves our product is aweso..."
In this segment, Varun discusses the necessity of continuous innovation in technology. He highlights that insights can depreciate quickly and emphasizes the need for companies to keep proving their value over time, using Nvidia as an example of a company that must innovate to maintain its competitive edge.
"everyone would be at mentioning everything, right? This almost reminded us of like the anti pattern of what Google and and these search engines were before like Google improved their product a lot whi..."
Varun elaborates on the concept of a competitive moat, explaining that it should be viewed as a verb rather than a noun. He stresses the importance of ongoing insights and execution to avoid stagnation, reinforcing the idea that companies must adapt and evolve to survive in a fast-paced tech landscape.
"like is that a very obvious decision now I think it's very obvious now it looks very obvious and this is where like one of the things that I think is is true for any startup is you have to keep provin..."
Varun reflects on how past experiences in GPU deployment and VMware optimization have contributed to Windsurf's current capabilities. He discusses the serendipitous nature of these experiences compounding into a competitive advantage, showcasing how previous knowledge can inform future innovations.
"Yeah. Something that could change with time right I also think for us and I tell the company this if we're not continuing to have insights and that's why I'm completely okay with a lot of our insights..."
In this segment, Varun critiques the conventional approach to retrieval-augmented generation (RAG) in AI startups. He argues for a more nuanced understanding of user needs and emphasizes that while vector databases are useful, they should not be the sole focus. Instead, he advocates for a comprehensive approach to delivering relevant answers.
"products and we work with OC companies. A lot of uh codegen tools use uh vector databases because we work with a lot of companies and that's was the standard approach how a lot of folks were building...."
Varun discusses the complexity of AI systems, particularly in the context of autonomous vehicles. He explains that while complexity is often seen as a drawback, it can be beneficial when paired with robust evaluation systems. This segment highlights the importance of balancing complexity with functionality in AI development.
"abstract syntax tree parsing and then on top of that using as you mentioned all the GPU infrastructure we have to take large chunks of the codebase and in real time rank it right as the query is comin..."
Varun shares insights on how Windsurf evaluates code changes using real-time user data and open-source projects. He explains the process of assessing intent and accuracy in code modifications, emphasizing the need for rigorous evaluation before introducing complexity into AI applications.
"a lot of the companies started off working on autonomous vehicles and the reason why that's kind of important is these are systems where you can't just yolo these systems, which is to say you build th..."
In this segment, Varun shares insights on how Windsurf balances data-driven evaluations with intuitive development. He highlights the challenges of evaluating user interactions in real-time and the importance of user feedback in refining the product, illustrating the dual approach of vibes and evaluations.
"a lot of complexity for any of these AI apps. I think you need to like make a rigorous hill that you can actually climb, right? Otherwise, you're just shooting in the dark, right? Why would we add the..."
Varun explains how power users leverage Windsurf for complex engineering tasks. He discusses the evolution of user expectations and the tool's ability to handle large codebases, emphasizing the need for precise intent when using AI to make code changes.
"lot of uh there's been a lot of chatter on the internet about vibes code is only for toy apps windsurf is actually being used for real production large code bases can you tell us about how the power u..."
Varun addresses the learning curve associated with using AI tools like Windsurf. He advises users to commit code frequently and be patient with the tool's learning process, highlighting the importance of iterative development and the potential for AI to enhance productivity.
"audience, how should a user that uses windsurf properly provide this intent so that the changes are more surgical? Because what you're saying with the agents creating all these broad changes I've seen..."
In this forward-looking segment, Varun discusses the future of AI in software development, predicting that AI will significantly enhance various stages of the development lifecycle. He emphasizes the rapid advancements in AI capabilities and the potential for transformative changes in coding practices.
"I think that maybe that's like the big big sort of uh tip there which is that uh you know you don't want to get in a situation where you've made 20 changes and on top of that made some changes yoursel..."
Varun explores how the integration of AI tools like Windsurf is changing the role of engineers. He suggests that engineers will increasingly focus on hypothesis testing and research-oriented tasks, leading to a shift in the skills and qualities sought in future hires.
"this for this product surface. An example is git has these things called work trees which is like you can have many work trees and versions of the repository all in your all in the same directory. Um ..."
Varun shares insights on the type of engineers Windsurf seeks to hire, emphasizing the importance of high agency and a willingness to take risks. He argues that startups should prioritize hiring individuals who are not just focused on routine coding but are eager to innovate and solve complex problems. This segment highlights the evolving expectations of engineers in the age of AI.
"type of engineer going forward like you're looking for different things Yeah, I think for engineers that we hire, we want to look for people with really high agency that are willing to be wrong uh and..."
Varun discusses the changing dynamics within engineering teams as AI tools become more prevalent. He notes that while AI can handle many tasks, there will always be a need for human engineers to tackle complex challenges. This segment explores the balance between leveraging AI for efficiency and maintaining the essential human element in software development.
"like workh horses to just kind of get certain things done. I feel like in the old days this was like building Android apps. Um it's like you hired someone to do it because there were very few people w..."
In this segment, Varun explains how Windsurf conducts interviews to assess candidates' problem-solving skills in the context of AI tools. He emphasizes the importance of evaluating candidates' ability to think critically without relying solely on AI assistance. This approach reflects the need for engineers to possess strong foundational skills even as AI becomes more integrated into the development process.
"interview people? Yeah, I think I think we have a fairly rigorous and high technical bar and that has a that's a combination of we give interviews that actually allow people to use the AI to kind of s..."
Varun addresses the challenges posed by AI in technical interviews, noting that AI tools can easily solve isolated problems. He discusses the need for interview questions that go beyond algorithmic challenges to assess deeper problem-solving abilities. This segment highlights the evolving landscape of technical hiring as AI capabilities advance.
"and it's it's it's you're totally right. There's very few problems now that something like an O4 Mini is not able to solve now. Um, right? I mean, if you look at competitive programming, it's just it'..."
Varun shares how Windsurf's interview process includes open-ended questions that encourage candidates to explore trade-offs and constraints. He emphasizes the importance of intellectual curiosity and the ability to navigate complex problems, showcasing how the hiring process adapts to the changing role of engineers in a world increasingly influenced by AI.
"are that are both system designy plus algorithms algorithms related. But these are questions that are fairly open-ended, right? Uh there may not be a correct answer. there are there are trade-offs tha..."
Varun discusses the ambitious mission of Windsurf to drastically reduce the time required to build technology and applications. He highlights the need for more engineers to tackle the high ceiling of possibilities in AI development. This segment underscores the ongoing evolution of the development experience as AI tools continue to enhance productivity.
"it's a herculan task. We need to start targeting more of the development experience. Right? Right now, we've helped a lot with the code writing process. Um and and maybe the the the navigation of code..."
In this segment, Varun envisions a future where software development becomes democratized, allowing anyone to build applications tailored to their needs. He discusses the potential for AI to create custom software solutions on demand, transforming the traditional role of developers. This forward-looking perspective highlights the shift towards a more inclusive approach to technology creation.
"prompts. The other thing that he came up with that we've been using all at YC internally is a new agent infrastructure that has direct access, read access to our system of record, our Postgress databa..."
Varun reveals that a significant portion of Windsurf's users are non-developers who utilize the platform without traditional coding skills. He discusses how these users engage with the tool and the potential for Windsurf to cater to a broader audience. This segment emphasizes the platform's accessibility and its role in empowering individuals to leverage technology without extensive programming knowledge.
"have for themselves. Do many people use windsurf who don't know how to write code at all? It's actually a large a large number of our users. Yeah. Interesting. Yeah. How did they end up getting into w..."
Varun shares insights on the competitive landscape of AI coding tools, particularly in relation to foundation models like OpenAI's. He emphasizes the need for Windsurf to continuously enhance its offerings to stay relevant as the baseline for AI capabilities evolves. This segment discusses the importance of providing additional value and the challenges of keeping pace with advancements in AI technology.
"care about it for your evals at least. Yeah. No, you need to you need to care about it for your evals. That maybe that's like the hard part for me that to to imagine for the pure non-developer product..."
Varun outlines potential opportunities for startups in the AI coding space, highlighting the need for specialized solutions in areas like software migrations and bug resolution. He emphasizes the economic value of these tasks and the lack of startups focusing on them. This segment provides valuable insights for entrepreneurs looking to innovate in the AI development landscape.
"this is not the most concerning is let's suppose that you were to take the foundation model and it's providing 90%. It's reducing the time it takes by 90%. That actually means if we can deliver one or..."
Reflecting on his journey, Varun offers advice to aspiring entrepreneurs, emphasizing the importance of being open to change and pivoting quickly. He encourages founders to treat pivots as a badge of honor and to embrace the courage to adapt their ideas. This segment serves as a motivational call to action for those looking to navigate the challenges of building a startup.
"thing I like about them is that there's like it's not just an opportunity for like two startups. Like each one of those is like a bucket that could have like a hundred large companies in. We actually ..."