AI Native Dev

💥 𝐕𝐒𝐂𝐨𝐝𝐞 𝐟𝐨𝐫𝐤𝐬 𝐥𝐢𝐤𝐞 𝐂𝐮𝐫𝐬𝐨𝐫 𝐚𝐫𝐞 𝐰𝐢𝐧𝐧𝐢𝐧𝐠 — 𝐬𝐨 𝐌𝐢𝐜𝐫𝐨𝐬𝐨𝐟𝐭 𝐬𝐭𝐚𝐫𝐭𝐞𝐝 𝐩𝐥𝐚𝐲𝐢𝐧𝐠 𝐝𝐞𝐟𝐞𝐧𝐬𝐞. Cursor hit $100M ARR by doing what VSCode extensions couldn’t: customising the IDE itself for AI-native workflows. Now? Some of the most-used Microsoft extensions are quietly breaking on forked IDEs. Cursor, Windsurf, and others are feeling it. No official bans. No loud policy changes. Just subtle friction… and devs are noticing. This isn’t just marketplace politics,  it’s about: - Why VSCode extensions hit a wall - How forks broke past those limits - Microsoft’s quiet response (and why it matters for your stack) 🧠 If you care about AI tooling, extension APIs, or the future of your dev environment, this one’s worth a read. 👇 Full post written by Dion Almaer - link in the comments #AINativeDev #Cursor #Windsurf #VSCode #copilot

🚨 𝐄𝐯𝐞𝐧𝐭 𝐀𝐧𝐧𝐨𝐮𝐧𝐜𝐞𝐦𝐞𝐧𝐭 🚨 The AI tooling space is moving fast and it’s easy to feel like you’re always one step behind. Join us for 𝐀𝐈 𝐍𝐚𝐭𝐢𝐯𝐞 𝐃𝐞𝐯 𝐔𝐧𝐩𝐥𝐮𝐠𝐠𝐞𝐝: 𝐈𝐧𝐬𝐢𝐠𝐡𝐭𝐬 𝐨𝐧 𝐭𝐡𝐞 𝐀𝐈 𝐃𝐞𝐯𝐞𝐥𝐨𝐩𝐞𝐫 𝐓𝐨𝐨𝐥𝐢𝐧𝐠 𝐋𝐚𝐧𝐝𝐬𝐜𝐚𝐩𝐞, a candid and interactive panel exploring how devs and tech leaders are navigating the ever-evolving ecosystem of AI-native tools. 🗓️ Date: April 29, 2025 ⏰ Time: 5:00 PM BST This session brings together Simon Maple, Amir Shevat and Patrick Debois, the creator of the AI Native Developer Tool Landscape, and influential voices from across the tooling stack to dig into: ✨ The current state of AI-native developer tools 🛠️ Insights from curating and building the tool landscape 🔍 How open collaboration can accelerate discovery and adoption 🚀 Practical takeaways to stay ahead of the curve 👉 Secure your spot and be part of the movement redefining developer productivity #AINativeDev #DeveloperTools #AIDev #DevOps #AIProductivity #Webinar #TechLeadership #PatrickDebois #AmirShevat #SimonMaple

𝐒𝐩𝐞𝐜-𝐜𝐞𝐧𝐭𝐫𝐢𝐜 𝐝𝐞𝐯. 𝐓𝐨𝐨𝐥 𝐜𝐡𝐚𝐨𝐬. 𝐀𝐮𝐭𝐨𝐧𝐨𝐦𝐨𝐮𝐬 𝐰𝐨𝐫𝐤𝐟𝐥𝐨𝐰𝐬. In todays episode of the AI Native Dev podcast, Guy Podjarny (Tessl) sits down with Tom Hulme (GV (Google Ventures)) to unpack what AI-native software development really looks like and where it’s heading. 🎧 𝐘𝐨𝐮’𝐥𝐥 𝐡𝐞𝐚𝐫: – Why “specs over code” could reshape how we build – What 170+ AI dev tools reveal about the current ecosystem – Where most tools fall short and why loyalty is low – What needs to change for autonomy to be real – How the role of the developer is already evolving If you’re building, experimenting, or just trying to stay ahead, this one’s worth your time. 📍Recorded live at GV (Google Ventures)’s GenAI Meetup in London. #AINativeDev #LLMs #DeveloperTools #GenAI #SpecCentric #SoftwareEngineering #Tessl #GVteam #DevWorkflows

🧠 𝐓𝐨 𝐌𝐂𝐏 𝐨𝐫 𝐭𝐨 𝐀2𝐀? 𝐓𝐡𝐚𝐭’𝐬 𝐭𝐡𝐞 𝐪𝐮𝐞𝐬𝐭𝐢𝐨𝐧. We have been waiting for Google to start riding the MCP wave. They joined the momentum on the first day of Google Cloud Next, while simultaneously announcing their own protocol: 𝐀2𝐀 (Agent-to-Agent) lets agents communicate and collaborate directly — not just as tool users, but as peers. They were positioned as complementary. But if you look closely, they sit at different levels of abstraction — and hint at very different developer mental models. - MCP is JSON-over-REST. It’s frictionless. It’s what function calling wants to be. - A2A is trying to define conversation-level behaviour between agents. It feels… more like WSDL (iykyk). In Dion Almaer’s latest blog post, he explores the tension and possibility between the two. It also links to some real gems: open source projects, diagrams, prompt injection warnings, and thoughts from folks like Simon Willison and Ray Cromwell. 👇 Read it here, and let me know what you think. Which would you build on?

🧠 𝐖𝐡𝐚𝐭 𝐡𝐚𝐩𝐩𝐞𝐧𝐬 𝐰𝐡𝐞𝐧 𝐭𝐡𝐞 “𝐨𝐫𝐢𝐠𝐢𝐧𝐚𝐥” 𝐀𝐈 𝐜𝐨𝐝𝐢𝐧𝐠 𝐭𝐨𝐨𝐥 𝐪𝐮𝐢𝐞𝐭𝐥𝐲 𝐥𝐞𝐯𝐞𝐥𝐬 𝐮𝐩? While everyone’s chasing shiny new editors and agent-first workflows, GitHub Copilot has been quietly evolving. And it’s not just playing catch-up—it’s improving how we interact with code. In Patrick Debois latest post, he explores: - How prompt management is moving from “copy-paste hacks” to real system-level design - The arrival of 𝐀𝐠𝐞𝐧𝐭 𝐌𝐨𝐝𝐞, and why it changes how devs think about planning vs. executing - What advancements like MCP (Model Context Protocol) signal about the future of IDEs as extensible AI platforms - Why we might be underestimating Copilot’s comeback - This isn’t about hype. It’s about the small but foundational changes that shift how we build software. Curious what GitHub’s been cooking? Link to blog in comments 👇 

🧠 “𝐋𝐋𝐌𝐬 𝐚𝐫𝐞 𝐢𝐧𝐬𝐚𝐧𝐞𝐥𝐲 𝐠𝐨𝐨𝐝 𝐚𝐭 𝐢𝐦𝐢𝐭𝐚𝐭𝐢𝐨𝐧—𝐬𝐨 𝐠𝐨𝐨𝐝 𝐢𝐭 𝐥𝐨𝐨𝐤𝐬 𝐥𝐢𝐤𝐞 𝐢𝐧𝐧𝐨𝐯𝐚𝐭𝐢𝐨𝐧.” — 𝐀𝐥𝐞𝐱 𝐊𝐨𝐦𝐨𝐫𝐨𝐬𝐤𝐞 In our latest AI Native Dev episode, Alex Komoroske unpacks what it really means to build in an AI-native world—and it’s not just about letting ChatGPT autocomplete your code. We’re talking about a new mental model for software development—one where AI isn’t an assistant, but a co-creator. 🎙️ In this episode: • Why situated software—small, scrappy tools—might become the new norm. • How specs become the source of truth, and LLMs translate them into code. • What developers get wrong about context length—and how it changes everything. • Why building microapps with LLMs feels like “coaxing” code into existence. • The real challenge ahead: stitching together microapps into something bigger. 🔥 Whether you’re an engineer, architect, or just AI-curious—this one will shift how you think. 📅 Plus, the 𝐀𝐈 𝐍𝐚𝐭𝐢𝐯𝐞 𝐃𝐞𝐯 𝐂𝐨𝐧 𝐒𝐩𝐫𝐢𝐧𝐠 𝐄𝐝𝐢𝐭𝐢𝐨𝐧 𝐢𝐬 𝐜𝐨𝐦𝐢𝐧𝐠 𝐌𝐚𝐲 13𝐭𝐡. #AIForDevs #SpecDrivenDev #Microapps #DeveloperTools #AINativeDev #LLMs #Tessl

“𝐀𝐧 𝐢𝐧𝐯𝐞𝐬𝐭𝐦𝐞𝐧𝐭 𝐢𝐧 𝐤𝐧𝐨𝐰𝐥𝐞𝐝𝐠𝐞 𝐩𝐚𝐲𝐬 𝐭𝐡𝐞 𝐛𝐞𝐬𝐭 𝐢𝐧𝐭𝐞𝐫𝐞𝐬𝐭.” - 𝑩𝒆𝒏𝒋𝒂𝒎𝒊𝒏 𝑭𝒓𝒂𝒏𝒌𝒍𝒊𝒏 And yet… Unstructured content is noise to most. To AI, it’s signal. In the final post of our 4-part series, we dig into a quiet revolution happening inside developer teams: → Every unstructured artifact - README, PR, Slack chat, terminal output, can become part of a living, AI-powered knowledge base. → AI helps capture the “ask around” knowledge inside your team’s heads and turn it into shared memory. → The best teams are designing systems that learn with them, not just serve them. Because in a world where tools are leveling the playing field, what is your real competitive advantage? What your team uniquely knows. About your systems. Your customers. Your edge cases. And how fast they can learn, adapt, and improve together, with AI. The future of engineering isn’t just shipping code. It’s building compound knowledge. 🔗 Read the final post in the series by Patrick Debois #AINativeDev #DevEx #EngineeringCulture #KnowledgeOps #AIxDev

In the age of AI-assisted development, a shift is underway. As delivery becomes automated and code generation accelerates, 𝐭𝐡𝐞 𝐝𝐞𝐯𝐞𝐥𝐨𝐩𝐞𝐫’𝐬 𝐫𝐨𝐥𝐞 𝐢𝐬 𝐬𝐡𝐢𝐟𝐭𝐢𝐧𝐠—𝐟𝐫𝐨𝐦 𝐞𝐱𝐞𝐜𝐮𝐭𝐢𝐧𝐠 𝐬𝐨𝐥𝐮𝐭𝐢𝐨𝐧𝐬 𝐭𝐨 𝐝𝐢𝐬𝐜𝐨𝐯𝐞𝐫𝐢𝐧𝐠 𝐭𝐡𝐞 𝐫𝐢𝐠𝐡𝐭 𝐩𝐫𝐨𝐛𝐥𝐞𝐦𝐬 𝐭𝐨 𝐬𝐨𝐥𝐯𝐞. In his latest piece, Patrick Debois (𝘺𝘦𝘴, 𝘵𝘩𝘦 𝘧𝘢𝘵𝘩𝘦𝘳 𝘰𝘧 𝘋𝘦𝘷𝘖𝘱𝘴) dives deep into this next pattern: 🔍 “𝐅𝐫𝐨𝐦 𝐃𝐞𝐥𝐢𝐯𝐞𝐫𝐲 𝐭𝐨 𝐃𝐢𝐬𝐜𝐨𝐯𝐞𝐫𝐲” It’s a rallying cry for developers to embrace curiosity, creativity, and co-creation. Here are some ✨ highlights: - 𝐕𝐢𝐛𝐞 𝐜𝐨𝐝𝐢𝐧𝐠 as a method of exploration, not perfection (inspired by Karpathy’s tweet on AI pair programming). - Using AI not just for triaging incidents, but generating potential fixes to test and learn faster. - Letting 𝐜𝐮𝐬𝐭𝐨𝐦𝐞𝐫𝐬 𝐛𝐞𝐜𝐨𝐦𝐞 𝐜𝐨-𝐜𝐫𝐞𝐚𝐭𝐨𝐫𝐬, personalising apps on the fly and feeding insights back to the team. Questioning the role of SaaS when AI can now mold services into exactly what users need, ushering in a new era of Service-as-Software. As Patrick puts it: “𝑾𝒆’𝒓𝒆 𝒔𝒘𝒂𝒑𝒑𝒊𝒏𝒈 𝒄𝒐𝒅𝒆 𝒇𝒐𝒓 𝒄𝒖𝒓𝒊𝒐𝒔𝒊𝒕𝒚.” Feels good, doesn’t it? Learning and discovering new things. 🔗 Read the full post → link in the comments What have you uncovered in your latest round of prototyping with AI? #AI #DevTools #DevEx #DevOps #ProductDiscovery #DeveloperExperience #Innovation #AINative #VibeCoding #PatrickDebois

What happens when building software feels more like snapping a Polaroid than shipping to prod? It’s not about writing perfect code. It’s about lowering the friction to create. Think AI-assisted dev, fast feedback loops, and building by feel, not fear. Some devs hate it. Some love it. But it’s here — and it’s changing how we code. Not every project needs tests, pipelines, and 17 design tokens. Sometimes you need to feel your way into the problem, ship something scrappy, and see what happens. Vibe coding is: • Building before you overthink. • Prototyping with AI + instinct. • Prioritizing momentum over perfection. It won’t replace deep engineering, but it will lower the floor so more ideas can ship. written by Dion Almaer, read the full thing ↓ #vibecoding #AINativeDev #softwareengineering #buildfast #devtools

What if we could bring that same philosophy to the way we build software? In his latest post, Patrick Debois dives deep into a second AI-native pattern—one that flips our traditional dev mindset. Instead of obsessing over how things are built, we zero in on what we want built and let AI handle the rest. 🧠🤖 Here’s what’s inside: - Programming in 𝐧𝐚𝐭𝐮𝐫𝐚𝐥 𝐥𝐚𝐧𝐠𝐮𝐚𝐠𝐞 instead of just syntax - Building and sharing 𝐩𝐫𝐨𝐦𝐩𝐭 𝐥𝐢𝐛𝐫𝐚𝐫𝐢𝐞𝐬 like modern standard libs - Turning 𝐭𝐞𝐬𝐭𝐬 𝐢𝐧𝐭𝐨 𝐞𝐱𝐞𝐜𝐮𝐭𝐚𝐛𝐥𝐞 𝐬𝐩𝐞𝐜𝐬 to align AI with intent - Prioritizing 𝐠𝐨𝐚𝐥𝐬 𝐨𝐯𝐞𝐫 𝐠𝐫𝐚𝐧𝐮𝐥𝐚𝐫 𝐥𝐨𝐠𝐢𝐜 for faster iteration and innovation This isn’t just theory, it’s a practical shift in how we design, collaborate, and ship software in the age of LLMs. 💬 Got thoughts or experiments of your own? Share them below. #AINative #PromptLibraries #DevPatterns #AIxDev #LLMEngineering #SpecDrivenDev #Tessl