AI Makerspace

⚒️ Introducing the LangSmith Agent Builder, a *real* agent builder, not just for building workflows. 🧑💻 Interesting, the entire no-code interface is completely different from previous Agent Builders, yet super familiar to everyone who uses LLMs! Join us live to check out the new tooling from LangChain, and to compare it the recent OpenAI Agent Builder we covered in October. 🚀 We’ll build, ship, and share agents with both Agent Builders and talk about what’s going on under the hood of the LangSmith Agent Builder  in the new and improved LangChain v1.0 ecosystem. Catch us here https://lnkd.in/ee3Te4gS

What we’re building 🏗️, shipping 🚢, and sharing 🚀 this week: How to operate like a manager, not a programmer, when deploying AI agents. LangSmith Agent Builder by LangChain is now in Public Beta! You can now build production-ready agents in < 5 minutes without writing a line of code - simply chat with the Agent Builder. Under the hood, the builder compiles to a stateful LangGraph and uses the Deep Agents harness. This gives you: ✅ Planning & Delegation ✅ Built-in Persistence ✅ Human-in-the-loop interrupts Real agents, high ceiling. Tune in tomorrow as Dr. Greg and The Wiz break down the Agent Builder. 🔗 Read the blog: https://lnkd.in/gRNYAexs 🔗 RSVP https://luma.com/adh3n7s3 #agent #builder #deepagent #state #langgraph #langchain

Building real-world AI systems takes more than model know-how – it requires engineering discipline, practical workflows, and the ability to ship reliable features fast. In this session from AI Makerspace, Dr. Greg Loughnane and Chris "The Wiz 🪄" Alexiuk break down what it actually looks like to design, build, and deploy modern AI applications. Whether you're moving from experimentation to production or leveling up your engineering foundation, this talk brings together hands-on demos, best practices, and patterns you can apply immediately. What you’ll learn: 🔹 How to structure modern AI engineering workflows 🔹 Patterns for building reliable LLM-powered applications 🔹 Practical tips from two industry educators who train AI engineers full-time 🔹 Where AI engineering is headed – and how to skill up fast 🎥 Dr. Greg Loughnane Loughnane, Co-Founder & CEO of AI Makerspace. 🎥 Chris "The Wiz 🪄" Alexiuk, Co-Founder & CTO of AI Makerspace and Developer Advocate at NVIDIA. Watch the video here: https://lnkd.in/dvn2y4YH #AIEngineering #LLM #GenAI #MachineLearning #AIMakerspace #NVIDIA #ODSC #DeveloperTools

Building real-world AI systems takes more than model know-how – it requires engineering discipline, practical workflows, and the ability to ship reliable features fast. In this session from AI Makerspace, Dr. Greg Loughnane and Chris "The Wiz 🪄" Alexiuk break down what it actually looks like to design, build, and deploy modern AI applications. Whether you're moving from experimentation to production or leveling up your engineering foundation, this talk brings together hands-on demos, best practices, and patterns you can apply immediately. What you’ll learn: 🔹 How to structure modern AI engineering workflows 🔹 Patterns for building reliable LLM-powered applications 🔹 Practical tips from two industry educators who train AI engineers full-time 🔹 Where AI engineering is headed – and how to skill up fast 🎥 Dr. Greg Loughnane, Co-Founder & CEO of AI Makerspace. 🎥 Chris "The Wiz 🪄" Alexiuk, Co-Founder & CTO of AI Makerspace and Developer Advocate at NVIDIA. Watch the video here:  https://lnkd.in/dvn2y4YH Want more sessions like this? Get your pass to ODSC AI East:  https://odsc.ai/east/ Join leading AI builders and see the latest in LLM engineering and tooling. #AIEngineering #LLM #GenAI #MachineLearning #AIMakerspace #NVIDIA #ODSC #DeveloperTools

🏗️ Transformation of the Week: Will Holt - From Idea to Industry Intelligence Layer As part of the AI Makerspace AI Engineer Bootcamp, Will Holt developed BuildOptima, an AI intelligence layer poised to revolutionize the $800B+ U.S. residential construction market. 🛑 𝐓𝐡𝐞 𝐏𝐫𝐨𝐛𝐥𝐞𝐦: Builders rely on expert opinion for multi-million-dollar decisions. This leads to slow inventory and tied-up capital. ✅ 𝐓𝐡𝐞 𝐒𝐨𝐥𝐮𝐭𝐢𝐨𝐧: BuildOptima delivers instant, subdivision-level intelligence. It replaces $5k–$20k comp studies with data-driven analysis to optimize home size and finish mix, guaranteeing a 5-10% margin swing. ⚙️ 𝐇𝐨𝐰 𝐢𝐭 𝐰𝐨𝐫𝐤𝐬: A full-stack AI model suite (Fast-Seller Classifier, DOM Regression, Pricing Model) feeds a Recommendation Engine, delivering a ranked build this next list and generating a lender-ready story via LLM. Code: https://lnkd.in/givArYic Slides: https://lnkd.in/gGs4retk As Will stated, 𝘉𝘶𝘪𝘭𝘥𝘖𝘱𝘵𝘪𝘮𝘢 𝘪𝘴 𝘢𝘯 𝘰𝘧𝘧𝘦𝘯𝘴𝘪𝘷𝘦 𝘱𝘢𝘳𝘵𝘯𝘦𝘳 - 𝘩𝘦𝘭𝘱𝘪𝘯𝘨 𝘣𝘶𝘪𝘭𝘥𝘦𝘳𝘴 𝘮𝘰𝘷𝘦 𝘧𝘢𝘴𝘵𝘦𝘳, 𝘧𝘦𝘦𝘭 𝘮𝘰𝘳𝘦 𝘤𝘰𝘯𝘧𝘪𝘥𝘦𝘯𝘵 𝘢𝘣𝘰𝘶𝘵 𝘵𝘩𝘦𝘪𝘳 𝘤𝘢𝘱𝘪𝘵𝘢𝘭, 𝘨𝘳𝘰𝘸 𝘮𝘢𝘳𝘨𝘪𝘯𝘴, 𝘢𝘯𝘥 𝘶𝘭𝘵𝘪𝘮𝘢𝘵𝘦𝘭𝘺 𝘣𝘶𝘪𝘭𝘥 𝘮𝘰𝘳𝘦 𝘩𝘰𝘮𝘦𝘴 𝘸𝘪𝘵𝘩 𝘧𝘦𝘸𝘦𝘳 𝘳𝘦𝘴𝘰𝘶𝘳𝘤𝘦𝘴 𝘢𝘯𝘥 𝘭𝘦𝘴𝘴 𝘴𝘵𝘳𝘦𝘴𝘴. 𝐂𝐮𝐫𝐢𝐨𝐮𝐬 𝐭𝐨 𝐥𝐞𝐚𝐫𝐧 𝐡𝐨𝐰 𝐭𝐨 𝐚𝐝𝐚𝐩𝐭 𝐀𝐈 𝐭𝐨 𝐲𝐨𝐮𝐫 𝐨𝐰𝐧 𝐝𝐨𝐦𝐚𝐢𝐧? 𝐉𝐨𝐢𝐧 𝐭𝐡𝐞 𝐩𝐫𝐨𝐠𝐫𝐚𝐦 𝐭𝐡𝐚𝐭 𝐟𝐢𝐭𝐬 𝐛𝐞𝐬𝐭 ✅ AI Engineer Onramp: Master the essentials and deploy 3 apps in 4 weeks. 🔗https://lnkd.in/gW7F-N5X ✅ AI Engineer Bootcamp: Go deep on RAG, evaluation, and agentic patterns. 🔗https://lnkd.in/g96iKq4f #ConstructionTech #PropTech #RealEstateTech #ResidentialConstruction #AI #MachineLearning #AIEngineering #LLM #DataScience #AIinConstruction OpenAI Qdrant Docker, Inc LangChain

THE BATTLE FOR PARSIMONY... I spent an absurd amount of time trying to teach Claude Code a simple lesson 🤦♂️🤖: Not every security problem needs a spaceship. 🚀🛑 All I wanted was to secure a FastMCP server. All Claude wanted was to re-architect OAuth2 flows, custom middleware, Kubernetes failover, and a Redis-backed token validator. 🛰️🔧 The real issue wasn’t that Claude didn’t know how to solve it. It’s that Claude couldn’t see that the simplest solution is often the right one. 🎯 So I stripped everything down. No app logic. No scaffolding. Just a tiny echo server and one Bearer token (API Key). 📦🔐 And suddenly… the AI stopped trying to build a distributed auth platform and finally saw the simple path forward. ✨ That’s the power of parsimony.   When the example is radically simple, the solution becomes clear — even for an AI that defaults to over‑engineering. 🧠⚙️ If you want the walkthrough that came out of this struggle (the simple, config‑first FastMCP auth path that Claude finally understood), it’s here: 👉 FastMCP Cloud Authentication Walkthrough:   https://lnkd.in/g6ncw86g Parsimony: 1, AI: 0.   (Claude still thinks we needed a service mesh. 😅) #AI #ClaudeCode #FastMCP #Parsimony #Overengineering #DeveloperHumor #AgenticAI AI Makerspace

Not everything needs a roadmap. Sometimes, you just need to vibe-code it. Over the weekend, I used Vercel’s v0 to spin up a self-help dashboard, pure vibe coding. No PRD. No Figma. Just flow state and the urge to build. It’s a lightweight client side React/Next.js app that lets users define and persist their top 3 daily focus tasks with localStorage. Built with componentized architecture, stateful UI, and minimal responsive design. Code: https://lnkd.in/gmgbKxiY What I learned: 1. Vibe Coding – Pros: - Rapid prototyping:  v0’s natural language prompts + component generation get you building in minutes. - Great for early validation:  early-stage concepts become clickable quickly. - Bridges product & engineering: early demos = faster alignment. 2. But… the Cons: - Beyond the happy path, real code takes over fast. - Testing, linting, structure? Still your job. - AI-generated UIs ≠ good UX, design debt creeps in fast. 3. Vibe Coding ≠ Low-Code ≠ AI-Assisted Dev As Dr. Greg Loughnane from AI Makerspace mentioned: “If all you can do is work in low-code environments, then you’re not using developer tools, and therefore, you’re not a developer.” That distinction matters - AI assistance augments. Vibe coding accelerates. But engineering still means knowing how to debug, refactor, version, ship. As a technologist and builder at heart, I am grateful to AI Makerspace for the nudge. It reminded me why I still write code: to think clearly, test fast, and lead better. Whether you are validating a side idea or leading product teams,  get your hands dirty. Tools change. Principles don’t. #vibecoding #aidevelopment #vercelv0 #reactjs #nextjs #engineeringleadership #buildinpublic #sideprojects #productivitytools #frontenddevelopment #aifuture #handsontools #aiengineer #aimakerspace #technicalleadership #weekendproject #selfhelpapps #uxuidesign #codeandbuild #minimaldesign

I have been posting about Cloudey already, but I realized I have not shared my demo yet, so here it is. This is the prototype of what I built for my AI Makerspace Demo Day presentation: an AI platform that gives non-technical teams instant cloud cost answers. No tickets to engineering. No waiting days for reports. The response validated there's a real need here. I'm documenting the whole process. If you want to see what it takes to go from bootcamp demo to real product, follow along. #BuildingInPublic #Cloudey #AIEngineering #CloudCosts