Patrick Collison

New course to bring you up to state-of-the-art at using AI to help you code: Build Apps with Windsurf's AI Coding Agents, built in partnership with Windsurf and taught by Anshul Ramachandran! AI-assisted IDEs (Integrated Development Environments) make developers’ workflows faster, more efficient, and much more fun. Agentic tools like Windsurf are more than just code autocomplete—they are collaborative coding agents that help you break down complex applications, iterate efficiently, and generate code that spans multiple files. Although a lot of coding assistants share the same underlying large language models for planning and reasoning, a major point of distinction is how they handle tools, keep track of context, and stay aligned with your intent as a developer. For instance, if you make modifications to a class definition in your code and make the same modifications to other classes in the same directory, you might tell the AI agent "Do the same thing in similar places in this directory." Here, tracking your intent means understanding that “the same thing" refers to that recent edit you just made, which must be followed by appropriate search and tool-calling to implement the changes. In this course, you'll learn the inner workings of coding agents, their strengths and limitations, and how to use Windsurf to quickly build several applications. In detail, you'll: - Build a mental model of how agents work by combining human-action tracking, tool integration, and context awareness to carry out an agentic coding workflow. - Learn the challenges of code search and discovery and how a multi-step retrieval approach helps coding agents address them. - Use Windsurf to analyze and understand a large, old codebase and update it to the latest versions of the frameworks and packages it uses. - Build a Wikipedia data analysis app that retrieves, parses, and analyzes word frequencies. - Enhance the performance of your Wikipedia analysis app by adding caching, and through this, also learn how to course-correct when the AI agent produces unexpected results. - Learn tips and tricks such as keyboard shortcuts, autocomplete, and @ mentions to quickly call on agentic capabilities. - Use image/multimodal capabilities of the AI agent to increase your development velocity; you'll see an example of uploading a mockup with sketched-out UI features, and ask the agent to use that to build new functionality to an app. By the end of this course, you’ll understand agentic coding in-depth and know how to use it to make your development process much faster, more efficient, and enjoyable. Please sign up here! https://lnkd.in/gaw3vq_u

2,293

106 Comments

Today we Lightspeed are excited to announce the industry-first formal partnership with 𝐒𝐤𝐲 𝐂𝐨𝐦𝐩𝐮𝐭𝐢𝐧𝐠 𝐋𝐚𝐛 to support the next generation of entrepreneurs & technical leaders at University of California, Berkeley. Many of the most consequential AI-native startups of our generation were founded by students, postdocs, faculty, and alums of UC Berkeley, including OpenAI, Databricks, Perplexity, Skild AI, ReflectionAI, Physical Intelligence, and more. Sky Computing Lab has followed in the footsteps of predecessors Berkeley RISE Lab & AMPLab to push technical research & commercialization across both AI/ML infrastructure and applications. Generational startups often begin within the walls of academic institutions like UC Berkeley, but the research funding landscape (in particular tied to NSF) is shifting, and the impact is already visible: fewer grants, slower timelines, and heightened urgency to commercialize. On behalf of Lightspeed, James Alcorn & I are honored to partner with Ion Stoica, Joseph Gonzalez, Dawn Song, and the Sky leadership team with a commitment of capital, network, and startup resources to ensure Berkeley's next wave of entrepreneurs has the platform and partners to build industry-defining companies. UC Berkeley College of Engineering UC Berkeley College of Computing, Data Science, and Society UC Berkeley Electrical Engineering & Computer Sciences (EECS) #UCBSkyComputing #BerkeleySky #UCBerkeley https://lnkd.in/etDTiBzC

257

18 Comments

The next frontier of AI isn't just about choosing the right model. It's about building moats through implementation. Across the Emergence Capital portfolio, companies like Bland and Federato are scaling faster by embedding Forward-Deployed Engineers. These engineers bridge the gap between AI and reality, ensuring real results in production. Read more from my colleagues Gordon and Kyle on why FDEs are the key to unlocking AI's full value. https://lnkd.in/gcsRRGX2

29

3 Comments

I’ve been working on the theory of why small, shippable factories are the perfect inversion to the centralized prefab factory model. It all collapsed into one line of math (credit to Gilles Retsin): Factory Efficiency = Usage × (Throughput × Product Value ) / ( CapEx + Deployment OpEx) The idea is that the closer you can drive Usage to 100 % while keeping the denominator tiny, the faster the flywheel spins.  A shippable microfactory tackles this by primarily focusing on a high Usage factor. Because the microfactory can be packed up afterwards and reused, its CapEx gets amortized over multiple projects, effectively raising its overall Usage across the year. If one project alone doesn’t fully occupy the factory, it can simply roll over to another job to maintain high Usage. Its CapEx is often an order of magnitude lower than a centralized factory (< $1M vs. $10-50M min), so the breakeven throughput is more achievable on a small pipeline of work. The real breakthrough here is portability. There is near-zero stranded capital and almost no idle time. To flesh this theory out, I wrote 3000 words on the topic for Brad Hargreaves and Thesis Driven. The article breaks down: - why “fixed factories, shipped goods” is being inverted to shipped factories, fixed goods - what a sub‑$1 M robotic cell does to CapEx per home versus a £45 M off‑site plant - early production/cost data on a shippable microfactory from Mollie Claypool, Gilles Retsin, Sam Baker and the Automated Architecture (AUAR) team - the hybrid reality of robots tackling the heavy structural components and human crews assembling faster and cheaper - technical constraints to pull this model off and future considerations on the optimal business model (own vs. rent) Major shoutout to Gilles Retsin for helping me pull this together. https://lnkd.in/eRTnbZPa

54

2 Comments

Despite all the big funding rounds and flashy demos in US robotics, K-Scale’s inability to raise more money should worry us. We're at risk of replaying the LLM story all over again in robotics: - Chinese companies are going open-source and collaborating across the value chain (from EV suppliers to downstream integrators) - most western teams are going full-stack proprietary, closed-source, all in-house Guess which robots the next wave of research labs and startups will actually be able to build on when they want to invent new algorithms or tackle unseen real-world use cases?

273

33 Comments

The AI Breakfast this Tuesday 6th May was on the theme of "Vibe Coding" and impact of LLM powered co-pilots and coding agents on software development. We had a great panel, with Jack Jackson founder of Recurse ML, Sean Ward co-founder of iGent AI and Jose Palazon the co-founder of Sagittal.ai and some illuminating but strong views were shared ( Yes a panelist did say that Vibe Coding is as useful as "Cancer" whilst on the other side of the argument another panellist did joke that in 5 years time most software developers may be flipping burgers) I think my main takeaways were: A: There is a big chasm difference between Software Developers/ Engineers using the likes of Cursor or Windsurf to autocomplete repetitive tasks and non technical people using prompting of LLM's and powerful distillation methods to try to reproduce other companies code bases or replicate others digital products - the consensus was beyond clever demo's and simple web sites the product of "vibe coding" fails to work when deployed in the real world and in a real IDE. How long will this be the case is up for hot debate B: Evaluation, Evaluation, Evaluation - How to understand, test, and gain trust and reassurance before and post deployment of a code base is increasingly the role of a SWE. LLM's can help you design and understand the software stack specification you want to build, some LLM's even have Taste ( an intuitive understanding of what could work and what will not that the best SWE have) but final judgement is still the human especially in regards to agile deployment C: "Pareto 80-20 rule" for using code-generators to develop deployable software only applies if you understand where the 80-20 line is - 80% of the journey to where? where and when do you exactly need the human SWE to take over? D: Those that use no-code natural language sites that expertly prompt LLM's to generate a simple code base and/or simple digital products like a website or mobile app because they cannot code will be locked into these platforms. Because coding will remain a black box to these users they will become dependent on using the function improving plug-ins offered by their no code platforms (like Lovable and Rock) and they in turn will probably become market places for no-code LLM powered applications. E: Agentic Dev Opps tools are about to take off. Lots of discussion on what's good and what is coming down the pipeline but there was consensus that LLM powered Dev Ops tools used by humans are going to be very different from those that feed into and are used by LLM powered Agents and this could become another big digital divide. Anyway it was an informative discussion followed by a very good breakfast. The next AI Breakfast on the 3rd June will be on ML Ops - lots and lots of innovation to explore there for AI First startups.

79

3 Comments

GPUs have gotten faster, and models have gotten larger. But storage is still stuck in the past. Modern AI workloads need instant access to huge datasets. Instead, teams waste time on slow retrieval, idle GPUs, and costly pipelines. Archil is changing this. It delivers object-storage scalability at block-storage speeds. The platform allows for seamless, high-performance access to massive datasets. By eliminating cold starts and accelerating throughput 30X, Archil enables teams to fully utilize their infrastructure and move faster on training, analytics, and AI deployment. Archil’s founder Hunter Leath brings rare technical depth and operating experience from building Amazon EFS and optimizing Netflix’s cloud performance. This is a team building the foundational layer that AI workloads desperately need. I’m thrilled to lead Archil’s seed at Felicis with Nancy Wang and support their mission to reinvent cloud storage for the AI era. https://lnkd.in/gWueXhzz

65

2 Comments

Producer Claude just got an upgrade! Anthropic AI just dropped Claude Sonnet 4.5, their cutting-edge new model. But what makes the new Claude better than the old Claude? 💻 Faster code refactoring + debugging ⚡ Better workflows for messy, real-world tasks 🛠️ Easily processes incomplete reqs + legacy code These companies are making their models 2X better every year, but can they keep up this relentless pace? Alex Wilhelm and I consider the possibilities in this exclusive TWiST clip

26

5 Comments

The Bitter Lesson is wrong. General methods that leverage massive computation will only outperform human-designed, domain-specific solutions if we have access to high-quality data. While we can probably use LLMs to generate synthetic data to improve model performance, we will still need human data to engender new capabilities and to improve user experiences. But existing tools/systems for collecting human data are falling short. Data labelling with unskilled labor and annotation with domain experts are fundamentally different and will require different approaches. Annotation requires us to find and screen experts who can perform increasingly challenging work. Collecting high-quality annotations faster will require us to design and develop new approaches to empower amateurs and enable experts to increase their throughput. Techniques from scalable oversight research can be applied to improve the throughput and quality of annotation but we also need to think through how to streamline annotation processes. I've discussed these themes and others in the first blogpost I've written in about a decade. Link to post in comments!

77

14 Comments

Elie Schoppik has spent his career at the intersection of education and engineering. He built early ed-tech tools that helped people actually retain knowledge, founded a Bay Area coding bootcamp that trained thousands of developers through practice and feedback, and now leads Technical Education at Anthropic. Despite all that experience, Elie says most teams are using AI wrong: they’re treating it like an all-knowing oracle instead of a tutor. This is why Elie designs prompts, examples, and workflows that treat AI as a teacher, instead of a shortcut. In our conversation, we discuss: • Why relying on AI for “fast answers” erodes judgment and verification • How to turn prompts into guided lessons • Why context engineering is the most underrated skill in 2025 • Practical exercises any team can run tomorrow to raise AI fluency • What leaders should steal from Anthropic’s approach to technical education  Episode is live now — full conversation linked in the comments.

35

2 Comments

While OpenAI signed $1.15 trillion in compute contracts through 2035, DeepSeek trained a frontier model for $6 million. This was 2025’s central question : are we building on bedrock or quicksand? The top 10 posts of 2025 examined some of these topics : Are we in a bubble echoing the telecom crash, or building the next internet? Do traditional exit paths still work when secondaries dominate & IPOs vanish? How do you design tools when the user is AI, not human? 2025 forced a reckoning with reality. 1. How AI Tools Differ from Human Tools (https://lnkd.in/d9qcnbyz) : I consolidated my 100+ AI tools into unified, parameter-rich interfaces based on Anthropic’s research. The counterintuitive finding : AI systems need complex tools with complete context, while humans need simple, chunked interfaces. Claude’s success rate approached 100% after the redesign. 2. Back to Text (https://lnkd.in/dWcy53fv) : How AI Might Reverse Web Design : I watched an open-source agent book flights by navigating airline websites, extracting data from visual chaos. If AI thrives on pure text, the future of the web might look exactly like it started : simple text, but for robots instead of humans. The better AI performs, the fewer websites we’ll visit. 3. Circular Financing (https://lnkd.in/d8-KMGZf) : Does Nvidia’s $110B Bet Echo the Telecom Bubble? : Nvidia’s vendor financing totals $110B in direct investments plus $15B+ in GPU-backed debt, 2.8x larger relative to revenue than Lucent’s exposure in 2000. But unlike the telecom bubble, Nvidia’s top customers generated $451B in operating cash flow in 2024. The merry-go-round has paying riders. Read the full post here : https://lnkd.in/d8Xw6tJt

43

24 Comments

Generative AI’s greatest potential lies in the physical world, where the ability to simulate, design, and iterate faster can reshape trillion-dollar sectors. That’s why we’re thrilled to announce that N47 is leading Luminary Cloud’s $72M Series B. The Luminary team is pioneering the Physics AI Model Factory, underpinned by a GPU-native physics simulation platform. This technology is designed from the ground up to unleash the performance of modern GPUs and generate the high-value datasets that power next-generation AI. Now, engineers can run hundreds of design iterations overnight instead of waiting days. AI models trained on physics-grounded data can suggest new aerodynamic designs or optimize energy systems in ways previously impossible, and companies across industries can accelerate their innovation cycles while reducing cost. With co-founders Juan Alonso and Pete Schlampp, Luminary has the technical depth and operational experience to define the new era of Physics AI. We’re proud to partner with them as they create the future of simulation and redefine what’s possible in the built world. Learn more about Luminary Cloud’s vision here: https://lnkd.in/gbjeWrvy 

119

6 Comments