🕰️ The Oral History of Windsurf (ft. Varun Mohan, Scott Wu, Jeff Wang, Kevin Hou, Anshul R)

We first had Nathan on to give us his RLHF deep dive when he was joining AI2, and now he’s back to help us catch up on the evolution to RLVR (Reinforcement Learning with Verifiable Rewards), first proposed in his Tulu 3 paper. While RLHF remains foundational, RLVR has emerged as a powerful approach for training models on tasks with clear success criteria and using verifiable, objective functions as reward signals—particularly useful in domains like math, code correctness, and instruction-following. Instead of relying solely on subjective human feedback, RLVR leverages deterministic signals to guide optimization, making it more scalable and potentially more reliable across many domains. However, he notes that RLVR is still rapidly evolving, especially regarding how it handles tool use and multi-step reasoning. We also discussed the Tulu model series, a family of instruction-tuned open models developed at AI2. Tulu is designed to be a reproducible, state-of-the-art post-training recipe for the open community. Unlike frontier labs like OpenAI or Anthropic, which rely on vast and often proprietary datasets, Tulu aims to distill and democratize best practices for instruction and preference tuning. We are impressed with how small eval suites, careful task selection, and transparent methodology can rival even the best proprietary models on specific benchmarks. One of the most fascinating threads is the challenge of incorporating tool use into RL frameworks. Lambert highlights that while you can prompt a model to use tools like search or code execution, getting the model to reliably learn when and how to use them through RL is much harder. This is compounded by the difficulty of designing reward functions that avoid overoptimization—where models learn to “game” the reward signal rather than solve the underlying task. This is particularly problematic in code generation, where models might reward hack unit tests by inserting pass statements instead of correct logic. As models become more agentic and are expected to plan, retrieve, and act across multiple tools, reward design becomes a critical bottleneck. Other topics covered: - The evolution from RLHF (Reinforcement Learning from Human Feedback) to RLVR (Reinforcement Learning from Verifiable Rewards) - The goals and technical architecture of the Tulu models, including the motivation to open-source post-training recipes - Challenges of tool use in RL: verifiability, reward design, and scaling across domains - Evaluation frameworks and the role of platforms like Chatbot Arena and emerging “arena”-style benchmarks - The strategic tension between hybrid reasoning models and unified reasoning models at the frontier - Planning, abstraction, and calibration in reasoning agents and why these concepts matter - The future of open-source AI models, including DeepSeek, OLMo, and the potential for an “American DeepSeek” - The importance of model personality, character tuning, and the model spec paradigm - Overoptimization in RL settings and how it manifests in different domains (control tasks, code, math) - Industry trends in inference-time scaling and model parallelism Finally, the episode closes with a vision for the future of open-source AI. Nathan has now written up his ambition to build an “American DeepSeek”—a fully open, end-to-end reasoning-capable model with transparent training data, tools, and infrastructure. He emphasizes that open-source AI is not just about weights; it’s about releasing recipes, evaluations, and methods that lower the barrier for everyone to build and understand cutting-edge systems. It would seem the Chapters 00:00:00 Welcome and Guest Introduction 00:01:18 Tulu, OVR, and the RLVR Journey 00:03:40 Industry Approaches to Post-Training and Preference Data 00:06:08 Understanding RLVR and Its Impact 00:06:18 Agents, Tool Use, and Training Environments 00:10:34 Open Data, Human Feedback, and Benchmarking 00:12:44 Chatbot Arena, Sycophancy, and Evaluation Platforms 00:15:42 RLHF vs RLVR: Books, Algorithms, and Future Directions 00:17:54 Frontier Models: Reasoning, Hybrid Models, and Data 00:22:11 Search, Retrieval, and Emerging Model Capabilities 00:29:23 Tool Use, Curriculum, and Model Training Challenges 00:38:06 Skills, Planning, and Abstraction in Agent Models 00:46:50 Parallelism, Verifiers, and Scaling Approaches 00:54:33 Overoptimization and Reward Design in RL 01:02:27 Open Models, Personalization, and the Model Spec 01:06:50 Open Model Ecosystem and Infrastructure 01:13:05 Meta, Hardware, and the Future of AI Competition 01:15:42 Building an Open DeepSeek and Closing Thoughts

This is a recap episode that ends with a short fresh interview on the future of Windsurf + Cognition with Jeff Wang and Scott Wu at the end. As the story of Windsurf as an independent company has come to a dramatic close with Google and Cognition, we’re taking this opportunity to look back at our coverage of Windsurf over the last 3 years. Here’s a brief timeline with related links. Jun 2021 - Exafunction founded Oct 2022 - Codeium pivot https://windsurf.com/blog/beta-launch-announcement Dec 2022 - “Copilot for X” https://www.latent.space/p/what-building-copilot-for-x-really Mar 2023 - Codeium first episode, LS episode 2 https://www.latent.space/p/varun-mohan July 2023 - “How to Make AI UX Your Moat" ****https://www.latent.space/p/ai-ux-moat Mar 2024 - Cognition Devin launch https://www.youtube.com/watch?v=fjHtjT7GO1c Jun 2024 - Scott @ AI Engineer https://www.youtube.com/watch?v=T7NWjoD_OuY Jun 2024 - Kevin @ AI Engineer https://www.youtube.com/watch?v=DuZXbinJ4Uc Nov 2024 - “Enterprise Infra Native” https://www.latent.space/p/enterprise Nov 2024 - Windsurf launch, LS Episode https://www.latent.space/p/windsurf Mar 2025 - Kevin Hou @ AI Engineer https://www.youtube.com/watch?v=bVNNvWq6dKo Jun 2025 - Scott @ AI Engineer https://www.youtube.com/watch?v=MI83buT_23o Jun 2025 - Kevin Hou @ AI Engineer https://www.youtube.com/watch?v=JVuNPL5QO8Q Jul 2025 - Jeff + Scott, CogSurf Episode ← new one, released here. We hope this serves as food for thought for students of history, and a reintroduction to the Latent Space extended universe and backlog, for those of you who are new. Welcome! Timestamps [00:02:07] Mar 2024 Codeium @ LS [00:52:36] Mar 2024 Devin Launch Video [00:54:28] Jun 2024 Codeium @ AIE SF [01:12:14] Jun 2024 Cognition @ AIE SF [01:30:53] Nov 2024 Windsurf Launch Video [01:37:16] Nov 2024 Windsurf Launch @ LS [02:43:10] Feb 2025 Windsurf @ AIE NYC [03:03:27] Jun 2025 Cognition @ AIE SF [03:18:50] June 2025 Windsurf @ AIE SF [03:34:23] July 2025 - Cognition + Windsurf Chapters 00:00:00 Mar 2024 Codeium @ LS 00:52:36 Mar 2024 Devin Launch Video 00:54:28 Jun 2024 Codeium @ AIE SF 01:12:14 Jun 2024 Cognition @ AIE SF 01:30:53 Nov 2024 Windsurf Launch Video 01:37:16 Nov 2024 Windsurf Launch @ LS 02:43:10 Feb 2025 Windsurf @ AIE NYC 03:03:27 Jun 2025 Cognition @ AIE SF 03:18:50 June 2025 Windsurf @ AIE SF 03:34:23 July 2025 - Cognition + Windsurf

ChatGPT handles 2.5B prompts/day and is on track to match Google's daily searches by end of 2026. AI agents don't browse like us—they crave queryable, chunkable data for tools like ChatGPT & Perplexity. A new industry is being born, some are calling it AI SEO, others GEO, but what is clear is that it drives amazing results. Businesses are seeing 2-4x higher conversion from visitors coming from AI compared to traditional search. Robert McCloy is the co-founder of Scrunch AI (https://scrunchai.com/), a fast growing company that helps brands and businesses re-write their content on the fly based on what agents are looking for. Chapters 00:00:00 Intro & Guest Introduction 00:01:30 The Genesis of Scrunch AI & AI Search Impact 00:06:02 AI Search Engines vs. Traditional SEO 00:06:28 Monitoring Prompts & The AI Search Stack 00:08:26 AI Training Data, Crawlers, and Content Strategy 00:12:33 AI Browsers and the Future of Web Consumption 00:16:06 Technical Mechanisms of AI Search & SEO Relevance 00:28:44 Personalization, Agent Experience, and Customer Journeys 00:30:44 Prompt Clusters, User Intent, and B2B Buying Patterns 00:36:06 Optimization Tactics: Prompt Injection, Content, and Pitfalls 00:40:37 Technical Content Delivery: JavaScript, Programmatic SEO, and LMS.txt 00:47:31 Case Studies & Conversion Optimization 00:51:36 Market Share & Platform Trends in AI Search 00:55:10 Wrap-Up & Future of AI-Driven Web

Akshay Agrawal joins us to talk about Marimo and their vision for the future of Python notebooks, and how it’s the perfect canvas for AI-driven data analysis. 0:00 Introduction 0:46 Overview of Marimo and Its Features 2:33 Origin Story and Motivation Behind Marimo 4:26 Demo: Classical Machine Learning with MNIST in Marimo 6:52 Notebook Compatibility and Conversion from Jupyter 7:42 Demo: Interactive Notebook with Custom UI and Layout 10:08 AI-Native Utilities and Code Generation with Language Models 11:36 Dependency Management and Integration with UV Package Manager 13:00 Demo: Data Annotation Workflow Using a PS5 Controller 15:51 Starting from Scratch: Blank Canvas AI Use Cases 18:27 Context Formatting for AI Code Generation 19:54 Chat Interface and Local/Remote Model Support 21:01 WebAssembly Support and MoLab Cloud-Hosted Notebooks 23:21 Future Plans and Breaking Out of Old Notebook Habits 25:40 Running Marimo Notebooks as Scripts or Data Apps 26:44 Exploring AI Agents and Community Contributions 26:56 Call to Action: How to Get Started and Contribute

Saoud Rizwan and Pash from Cline joined us to talk about why fast apply models got bitter lesson'd, how they pioneered the plan + act paradigm for coding, and why non-technical people use IDEs to do marketing and generate slides. Full writeup: https://www.latent.space/p/cline X: https://x.com/latentspacepod Chapters: 00:00 - Introductions 01:35 - Plan and Act Paradigm 05:37 - Model Evaluation and Early Development of Cline 08:14 - Use Cases of Cline Beyond Coding 09:09 - Why Cline is a VS Code Extension and Not a Fork 12:07 - Economic Value of Programming Agents 16:07 - Early Adoption for MCPs 19:35 - Local vs Remote MCP Servers 22:10 - Anthropic's Role in MCP Registry 22:49 - Most Popular MCPs and Their Use Cases 25:26 - Challenges and Future of MCP Monetization 27:32 - Security and Trust Issues with MCPs 28:56 - Alternative History Without MCP 29:43 - Market Positioning of Coding Agents and IDE Integration Matrix 32:57 - Visibility and Autonomy in Coding Agents 35:21 - Evolving Definition of Complexity in Programming Tasks 38:16 - Forks of Cline and Open Source Regrets 40:07 - Simplicity vs Complexity in Agent Design 46:33 - How Fast Apply Got Bitter Lesson'd 49:12 - Cline's Business Model and Bring-Your-Own-API-Key Approach 54:18 - Integration with OpenRouter and Enterprise Infrastructure 55:32 - Impact of Declining Model Costs 57:48 - Background Agents and Multi-Agent Systems 1:00:42 - Vision and Multi-Modalities 1:01:07 - State of Context Engineering 1:07:37 - Memory Systems in Coding Agents 1:10:14 - Standardizing Rules Files Across Agent Tools 1:11:16 - Cline's Personality and Anthropomorphization 1:12:55 - Hiring at Cline and Team Culture Chapters 00:00:00 Introduction and Guest Intros 00:00:29 What is Klein? Product Overview 00:01:42 Plan and Act Paradigm 00:05:22 Model Evolution and Building Klein 00:07:40 Beyond Coding: Klein as a General Agent 00:09:12 Why Focus on VS Code Extension? 00:11:26 The Future of Programming and Agentic Paradigm 00:12:34 Economic Value: Programming vs. Other Use Cases 00:16:04 MCP Ecosystem: Growth and Marketplace 00:21:30 Security, Discoverability, and Trust in MCPs 00:22:55 Popular MCPs and Workflow Automation 00:25:30 Monetization and Payments for MCPs 00:37:53 Competition, Forks, and Open Source Philosophy 00:40:39 RAG, Fast Apply, and Agentic Simplicity 00:50:11 Business Model and Enterprise Adoption 00:57:04 Background Agents, Multi-Agent Systems, and CLI 01:00:41 Context Engineering and Memory 01:12:39 Team, Culture, and Closing Thoughts

Speak (https://speak.com) may not be very well known to native English speakers, but they have come from a slow start in 2016 to emerge as one of the favorite partners of OpenAI, with their Startup Fund leading and joining their Series B and C as one of the new AI-native unicorns, noting that “Speak has the potential to revolutionize not just language learning, but education broadly”. Today we speak with Speak’s CTO, Andrew Hsu, on the journey of building the “3rd generation” of language learning software (with Rosetta Stone being Gen 1, and Duolingo being Gen 2). Speak’s premise is that speech and language models can now do what was previously only possible with human tutors—provide fluent, responsive, and adaptive instruction—and this belief has shaped its product and company strategy since its early days. https://www.linkedin.com/in/adhsu/ https://speak.com One of the most interesting strategic decisions discussed in the episode is Speak’s early focus on South Korea. While counterintuitive for a San Francisco-based startup, the decision was influenced by a combination of market opportunity and founder proximity via a Korean first employee. South Korea’s intense demand for English fluency and a highly competitive education market made it a proving ground for a deeply AI-native product. By succeeding in a market saturated with human-based education solutions, Speak validated its model and built strong product-market fit before expanding to other Asian markets and eventually, globally. The arrival of Whisper and GPT-based LLMs in 2022 marked a turning point for Speak. Suddenly, capabilities that were once theoretical—real-time feedback, semantic understanding, conversational memory—became technically feasible. Speak didn’t pivot, but rather evolved into its second phase: from a supplemental practice tool to a full-featured language tutor. This transition required significant engineering work, including building custom ASR models, managing latency, and integrating real-time APIs for interactive lessons. It also unlocked the possibility of developing voice-first, immersive roleplay experiences and a roadmap to real-time conversational fluency. To scale globally and support many languages, Speak is investing heavily in AI-generated curriculum and content. Instead of manually scripting all lessons, they are building agents and pipelines that can scaffold curriculum, generate lesson content, and adapt pedagogically to the learner. This ties into one of Speak’s most ambitious goals: creating a knowledge graph that captures what a learner knows and can do in a target language, and then adapting the course path accordingly. This level-adjusting tutor model aims to personalize learning at scale and could eventually be applied beyond language learning to any educational domain. Finally, the conversation touches on the broader implications of AI-powered education and the slow real-world adoption of transformative AI technologies. Despite the capabilities of GPT-4 and others, most people’s daily lives haven’t changed dramatically. Speak sees itself as part of the generation of startups that will translate AI’s raw power into tangible consumer value. The company is also a testament to long-term conviction—founded in 2016, it weathered years of slow growth before AI caught up to its vision. Now, with over $50M ARR, a growing B2B arm, and plans to expand across languages and learning domains, Speak represents what AI-native education could look like in the next decade. Chapters 00:00:00 Introductions & Thiel Fellowship Origins 00:02:13 Genesis of Speak: Early Vision & Market Focus 00:03:44 Building the Product: Iterations and Lessons Learned 00:10:59 AI’s Role in Language Learning 00:13:49 Scaling Globally & B2B Expansion 00:16:30 Why Korea? Localizing for Success 00:19:08 Content Creation, The Speak Method, and Engineering Culture 00:23:31 The Impact of Whisper and LLM Advances 00:29:08 AI-Generated Content & Measuring Fluency 00:35:30 Personalization, Dialects, and Pronunciation 00:39:38 Immersive Learning, Multimodality, and Real-Time Voice 00:50:02 Engineering Challenges & Company Culture 00:53:20 Beyond Languages: B2B, Knowledge Graphs, and Broader Learning 00:57:32 Fun Stories, Lessons, and Reflections 01:02:03 Final Thoughts: The Future of AI Learning & Slow Takeoff

When the first video diffusion models started emerging, they were little more than just “moving pictures” - still frames extended a few seconds in either direction in time. There was a ton of excitement about OpenAI’s Sora on release through 2024, but so far only Sora-lite has been widely released. Meanwhile, other good videogen models like Genmo Mochi, Pika, MiniMax T2V, Tencent Hunyuan Video, and Kuaishou’s Kling have emerged, but the reigning king this year seems to be Google’s Veo 3, which for the first time has added native audio generation into their model capabilities, eliminating the need for a whole class of lipsynching tooling and SFX editing. The rise of Veo 3 unlocks a whole new category of AI Video creators that many of our audience may not have been exposed to, but is undeniably effective and important particularly in the “kids” and “brainrot” segments of the global consumer internet platforms like Tiktok, YouTube and Instagram. By far the best documentarians of these trends for laypeople are Olivia and Justine Moore, both partners at a16z, who not only collate the best examples from all over the web, but dabble in video creation themselves to put theory into practice. We’ve been thinking of dabbling in AI brainrot on a secondary channel for Latent Space, so we wanted to get the braindump from the Moore twins on how to make a Latent Space Brainrot channel. Jump on in! Chapters 00:00:00 Introductions & Guest Welcome 00:00:49 The Rise of Generative Media 00:02:24 AI Video Trends: Italian Brain Rot & Viral Characters 00:05:00 Following Trends & Creating AI Content 00:07:17 Hands-On with AI Video Creation 00:18:36 Monetization & Business of AI Content 00:23:34 Platforms, Models, and the Creator Stack 00:37:22 Native Content vs. Clipping & Going Viral 00:41:52 Prompt Theory & Meta-Trends in AI Creativity 00:47:42 Professional, Commercial, and Platform-Specific AI Video 00:48:57 Wrap-Up & Final Thoughts

Our last AI PhD grad student feature was Shunyu Yao, who happened to focus on Language Agents for his thesis and immediately went to work on them for OpenAI. Our pick this year is Jack Morris, who bucks the “hot” trends by -not- working on agents, benchmarks, or VS Code forks, but is rather known for his work on the information theoretic understanding of LLMs, starting from embedding models and latent space representations (always close to our heart). Jack is an unusual combination of doing underrated research but somehow still being to explain them well to a mass audience, so we felt this was a good opportunity to do a different kind of episode going through the greatest hits of a high profile AI PhD, and relate them to questions from AI Engineering. Papers and References made AI grad school: https://x.com/jxmnop/status/1933884519557353716A new type of information theory: https://x.com/jxmnop/status/1904238408899101014EmbeddingsText Embeddings Reveal (Almost) As Much As Text: https://arxiv.org/abs/2310.06816Contextual document embeddings https://arxiv.org/abs/2410.02525Harnessing the Universal Geometry of Embeddings: https://arxiv.org/abs/2505.12540Language modelsGPT-style language models memorize 3.6 bits per param: https://x.com/jxmnop/status/1929903028372459909Approximating Language Model Training Data from Weights: https://arxiv.org/abs/2506.15553https://x.com/jxmnop/status/1936044666371146076LLM Inversion"There Are No New Ideas In AI.... Only New Datasets"https://x.com/jxmnop/status/1910087098570338756https://blog.jxmo.io/p/there-are-no-new-ideas-in-ai-onlymisc reference: https://junyanz.github.io/CycleGAN/ — for others hiring AI PhDs, Jack also wanted to shout out his coauthor Zach Nussbaum, his coauthor on Nomic Embed: Training a Reproducible Long Context Text Embedder.

Solving Poker and Diplomacy, Debating RL+Reasoning with Ilya, what's *wrong* with the System 1/2 analogy, and where Test-Time Compute hits a wall Timestamps 00:00 Intro – Diplomacy, Cicero & World Championship 02:00 Reverse Centaur: How AI Improved Noam’s Human Play 05:00 Turing Test Failures in Chat: Hallucinations & Steerability 07:30 Reasoning Models & Fast vs. Slow Thinking Paradigm 11:00 System 1 vs. System 2 in Visual Tasks (GeoGuessr, Tic-Tac-Toe) 14:00 The Deep Research Existence Proof for Unverifiable Domains 17:30 Harnesses, Tool Use, and Fragility in AI Agents 21:00 The Case Against Over-Reliance on Scaffolds and Routers 24:00 Reinforcement Fine-Tuning and Long-Term Model Adaptability 28:00 Ilya’s Bet on Reasoning and the O-Series Breakthrough 34:00 Noam’s Dev Stack: Codex, Windsurf & AGI Moments 38:00 Building Better AI Developers: Memory, Reuse, and PR Reviews 41:00 Multi-Agent Intelligence and the “AI Civilization” Hypothesis 44:30 Implicit World Models and Theory of Mind Through Scaling 48:00 Why Self-Play Breaks Down Beyond Go and Chess 54:00 Designing Better Benchmarks for Fuzzy Tasks 57:30 The Real Limits of Test-Time Compute: Cost vs. Time 1:00:30 Data Efficiency Gaps Between Humans and LLMs 1:03:00 Training Pipeline: Pretraining, Midtraining, Posttraining 1:05:00 Games as Research Proving Grounds: Poker, MTG, Stratego 1:10:00 Closing Thoughts – Five-Year View and Open Research Directions Chapters 00:00:00 Intro & Guest Welcome 00:00:33 Diplomacy AI & Cicero Insights 00:03:49 AI Safety, Language Models, and Steerability 00:05:23 O Series Models: Progress and Benchmarks 00:08:53 Reasoning Paradigm: Thinking Fast and Slow in AI 00:14:02 Design Questions: Harnesses, Tools, and Test Time Compute 00:20:32 Reinforcement Fine-tuning & Model Specialization 00:21:52 The Rise of Reasoning Models at OpenAI 00:29:33 Data Efficiency in Machine Learning 00:33:21 Coding & AI: Codex, Workflows, and Developer Experience 00:41:38 Multi-Agent AI: Collaboration, Competition, and Civilization 00:45:14 Poker, Diplomacy & Exploitative vs. Optimal AI Strategy 00:52:11 World Models, Multi-Agent Learning, and Self-Play 00:58:50 Generative Media: Image & Video Models 01:00:44 Robotics: Humanoids, Iteration Speed, and Embodiment 01:04:25 Rapid Fire: Research Practices, Benchmarks, and AI Progress 01:14:19 Games, Imperfect Information, and AI Research Directions

Chris Lattner of Modular (https://modular.com) joined us (again!) to talk about how they are breaking the CUDA monopoly, what it took to match NVIDIA performance with AMD, and how they are building a company of "elite nerds". X: https://x.com/latentspacepod Substack: https://latent.space 00:00:00 Introductions 00:00:12 Overview of Modular and the Shape of Compute 00:02:27 Modular’s R&D Phase 00:06:55 From CPU Optimization to GPU Support 00:11:14 MAX: Modular’s Inference Framework 00:12:52 Mojo Programming Language 00:18:25 MAX Architecture: From Mojo to Cluster-Scale Inference 00:29:16 Open Source Contributions and Community Involvement 00:32:25 Modular's Differentiation from VLLM and SGLang 00:41:37 Modular’s Business Model and Monetization Strategy 00:53:17 DeepSeek’s Impact and Low-Level GPU Programming 01:00:00 Inference Time Compute and Reasoning Models 01:02:31 Personal Reflections on Leading Modular 01:08:27 Daily Routine and Time Management as a Founder 01:13:24 Using AI Coding Tools and Staying Current with Research 01:14:47 Personal Projects and Work-Life Balance 01:17:05 Hiring, Open Source, and Community Engagement