Newsletter #64 – Podcast with  Hrishi Olickel, founder of SouthbridgeAI, relocating to Spain, AI Agents with JavaScript

For those of you who celebrated it, I hope you had a fantastic Easter break!

I’ve started a new job and relocated with my family to the south of Spain, in the Malaga/Gibraltar area. I’ll share more about this in the coming weeks, but in the meantime, feel free to reach out if you’re in the area.

Now, let’s get to the newsletter:

• 🎙️ I’m excited to share my interview with Hrishi Olickel, founder of SouthbridgeAI. We talked about his journey into AI Engineering, his work with multi-modal models, and his approach to learning. It's one of my favorite episodes. Enjoy!
  

• 🎓 Education has always been a huge part of my life. What the people at Alpha School are doing is absolutely incredible. While I still think the human interaction is essential for a good education, I cannot obtain myself Of getting overexcited about the future of AI & education.

• 🎧 This interview between Tim Ferris and Chris Williamson was the only 3-hour talk I recently had the patience to listen to, and it was worth every minute. If you're interested in understanding a sustainable personal operating system, give it a try. I especially liked Tim's focus on doing things that energize him.

• 📦 It’s safer to think of an LLM as a giant set of tables filled with numbers. An LLM consists of about 500 lines of code and countless gigabytes of raw data. These models are more like massive number-crunching machines than traditional if-then-else software.

• 📘 Just wanted to remind the ones who bought the “Building Al Agents with LangGraph.js” book that I’ve added 2 brand new chapters to it. One chapter focuses on multimodal AI Agents (using images and audio), while the other focuses on the ReAct architecture for AI Agents. You can download the updated version from the store you bought it from. Just reach out if you need help.

• 🐤 A personal tip on using Twitter. Create a list of people you want to follow or engage with directly. Relying solely on Twitter's algorithm feed is, unfortunately, becoming a time-waster.

• 📦 The MCP Model Context Protocol is taking the world by storm. I’ve made this short article & an example of how to build a client-server MCP app using JavaScript from scratch.

• 🎓 After a long wait, I’m finally launching an online learning community for readers of my LangGraph and LangChain books. I’ve used Slack for years, but Discord seems to be the new standard. Vote here or reply to this email to share your preference: Slack or Discord.

• 🗂️ Modern RAG is likely the most economically valuable form of Gen AI. Cursor or Claude Code are basically RAG apps. One of the real performance gains Cursor has made is using embeddings to figure out where your cursor should jump into for the next tab action. Jason Liu gave a great talk about this.

• 💡 I also recorded a podcast with Sam Bhagwat, founder of mastra.ai and Gatsby.js. A key takeaway: over half of the Winter 2025 Y Combinator batch is building AI Agent apps. Y Combinator knows what they’re doing, and this makes me even more bullish on AI.

• 🚀 And speaking of Sam, I really like this idea of Python being for LLM training, but not for AI Agents. I think that JavaScript is the language for web developer builders, and shortly, we will have a big increase in the JS AI tooling & frameworks.

• 🤖 This is an example of a real production prompt used for an AI agent. Check out the high level of complexity and detail involved. The prompt comes from the Bolt.new app. As with code, LLM prompts are constantly updated and evolving.

• ⛰️ Relocating near Gibraltar comes with its advantages, like a home office with an amazing view:

And that’s a wrap for this edition of the JS-Craft newsletter! Be kind, keep coding, and keep learning!

See you the next time!