Agent Meetup Oslo - Vol. 5

What is it? Our seasonal hackathon for 100 select developers. Come as you are, create something new, and impress some cool people. How does it work? We help you form teams of max 3 devs Build something people want in 10 hours Deliver a demo to your new friends in 2 minutes What should I expect? Get started with loops and the basics of Loop Engineering. Ship a 0 to 1 application or feature Compete to impress excellent developers ____________HOW TO WIN_______________ Who will jury? Markus Skagemo, CTO @ Crunched (YC F25) Stefan Mandaric, CTO @ Moritz (YC W26) Sven Myhre, CTO @ Ara (YC P26) Peter Flo, Chief Scientist @ Audun (YC S26) Bob Deforce, Startups @ Anthropic William Imoh, Team @ AI Mad Lab What criteria do they use? How real is the problem the project solves? 50% How well are the demos implemented? 50% How do I give a great demo? Share your team name Our solution helps [user X] solve [task Y] by [solution Z] Demo how a user would solve the task using the solution ____________PROGRAM_______________ What does our day look like? 09:30: Doors open! Best advise: be early 10:00am: Introduction to task 10:30am: Create a team 11:00am: Coding begins! 12:00pm: Free lunch 5:00pm: Free dinner 7:00pm: Live demos 7:30pm: Winners announced 8:00pm: Doors close Where do I go? Startuplab is located in Oslo Science Park Metro is in development, plan accordingly. Take a tram to Forskningsparken, bus to Gaustad, or a Voi from Majorstuen. Enter by main entrance (it's not the garage on Google Maps) _____________HOW TO PREPARE______________ Arrive with a couple ideas Everyone has equal opportunity to prepare. Recommended resources Models: Anthropic Hosting: Github MCP Deployment: Vercel MCP Database: Supabase MCP Sandboxes: Daytona MCP Compute: Runpod MCP Practicals Bring your laptop, charger, headset Use Guest Wifi. No password required Please leave rooms super clean _______________THANK YOU________________ Made possible by AI Mad Lab Startuplab

Welcome to Claude Fable 5.1 Build Day Oslo! Fable 5.1 is here, and we're getting the Oslo Claude community together to spend an afternoon pushing it as far as it goes. Bring your laptop and an idea, and we'll build and have some fun. Everything gets made in the room, so the only prep is to have Claude Code set up on your laptop before you arrive. Who this is for: Everyone. Developers, builders, and everyday Claude Code users. Whether you installed it yesterday or have been running agents in loops for months, you're welcome. Pick a track: Delight — build something surprising, beautiful, or just plain fun. The bar: you couldn't have made it this good, this fast, before. Breakthrough — build something that wasn't possible before. Take a problem the last model couldn't handle and see how far this one gets. Everyday — build something that fixes a real annoyance in your actual life or work. Teams: Go solo or work together in a group. Teams form on the day, nothing to arrange in advance. If you show up alone, you won't stay that way for long. Agenda: 17:30 — Doors open, food served 18:00 — Kickoff: intro and a few quick Fable 5.1 demos 18:30 — Build time (2 hours, pick your track and go) 20:30 — Participant demos 21:00 — Wrap up We have limited seats for this event, so grab yours early. The best time to build is now. 🚀

Join GDG Cloud Oslo for an exciting, hands-on hack-along focused on running Large Language Models (LLMs) securely in your local environment using Google's open weights model, Gemma. Dive into the mechanics of local LLMs, exploring essential practices to ensure security, predictability, and robustness. This workshop will provide architectural patterns and Java-based code examples, but feel free to use your preferred language to implement these security measures. To kickstart the workshop, we have prepared complete architectural patterns and code examples implemented in Java. However, this is a polyglot-friendly event! We highly welcome and encourage you to take these blueprints and implement the same security and predictability patterns in Python, Go, TypeScript, Rust, or any language of your choice. 🚀 Agenda Part 1: Gemma Intro & The Anatomy of Local LLMs Get up to speed with Google's Gemma model family. We'll discuss why running models locally is a game-changer, how the architecture works, and why local execution helps keep your data private. Part 2: The Core Challenge – Security & Predictability Using practical concepts and real-world vulnerabilities (inspired by OWASP Top 10 for LLMs), we will discuss why AI models can sometimes behave unpredictably. Learn how to tame non-deterministic outputs, avoid injection risks, and configure boundaries to ensure predictable behavior. Part 3: Hands-On Hack-Along! 💻 Bring your laptops! We will provide ready-to-run Java architectural patterns to guide your implementation. You can dive straight into the Java codebase or use it as a reference blueprint to rebuild the same secure LLM pipelines in your favorite programming language. 🎒 What to Bring Your laptop and charger. Your IDE of choice configured for your preferred language (Java, Python, Node, etc.). A curious mindset ready to code and break things safely! (Optional) Pre-install Ollama or Docker if you want a head start on running Gemma locally. Food and drinks w