Welcome to the Open Hack 👋
Welcome! Today is all about rolling up your sleeves, building something real, and having fun while doing it. 🎯
Open Hack is a challenge-driven, hands-on experience where you’ll collaborate with other developers to create AI-powered solutions using tools like Microsoft Foundry. This is not a lecture or a step-by-step lab—you’ll learn by building and solving real problems.
Purpose of the Day
The goal is:
- Learn by doing: ship a working solution through small, focused levels.
- Collaborate openly: ask questions early, help each other, and keep momentum.
By the end of the day, you’ll walk away with practical AI development experience and a clearer idea of how to apply these patterns in your own projects.
What to Expect
- Hands-on development from the start
- Progressively more advanced AI challenges
- Support from Microsoft engineers, AI experts, and on-site coaches
- A relaxed, collaborative environment with good food and great conversations
The Challenge
Today you’ll build an AI agent capable of understanding user requests and completing a task end-to-end.
The scenario for this Open Hack is pizza ordering 🍕:
- The agent interprets a user’s order
- Asks clarifying questions when needed
- Places the order using the provided services
The challenge is broken into small, incremental levels, each building on the previous one—so progress is steady and achievable.
🕹 The Levels
Each level builds on the previous one. Follow them in order to complete the full challenge.
1️⃣ Interact with Microsoft Foundry
Set up your environment, create a Microsoft Foundry resource, and deploy your first model.
Teaser: You’ll prove your setup works by chatting with a model in the playground.
2️⃣ Create Your First Agent
Bring your agent to life with a simple “hello world” response.
Teaser: Your agent will appear in Microsoft Foundry and greet you when called.
3️⃣ Add Instructions and Persistent Memory
Give your agent a personality, teach it to remember names, and keep it focused on pizza-only conversations.
Teaser: Your agent starts feeling less like a bot and more like a real pizza assistant.
4️⃣ Work with Store Information
Connect your agent to real Contoso Pizza store data so it can answer questions and ensure orders are tied to a specific location.
Teaser: Your agent learns where to send the pizza before confirming any order.
5️⃣ Add a Pizza Estimation Function
Make your agent smart enough to calculate how much pizza a group needs based on appetite and party size.
Teaser: “We have 8 hungry people.” → “You’ll probably need 4 large pizzas.”
6️⃣ Connect to the Pizza MCP Server
Move from direct API calls to a flexible MCP server connection. Your agent can now take, track, and even cancel real pizza orders.
Teaser: This is the big one—your agent becomes a true pizza-ordering assistant connected to the real system.
Coaches & Collaboration
You’re not on your own:
- Coaches are in the room - grab one anytime
- Talk to your neighbours: quick peer discussions solve most issues fast
- If you’re stuck, raise your hand
We’re all here to learn together.
Ready to Start?
Everything is set-let’s move on to the challenge and start building. 🚀