The Frontier Hack is a hands-on experience focused on exploring how AI agents are designed, tested, and extended using real tools and realistic constraints.
You’ll work individually or with your group to build and reason about agent behavior — from grounding agents in data, to shaping how they interpret intent, to thinking about how multiple agents and workflows come together as a system.
The emphasis is on learning by doing and comparing approaches, not on arriving at a single “correct” solution.
Develop practical instincts for designing agent-based solutions — including how decisions about data, reasoning, and orchestration affect real-world outcomes.
The Frontier Hack is organized into three zones, each exploring a different layer of agent design.
You can complete:
- all three zones
- just one or two zones
- the zones in any order
It’s entirely up to you and what you want to prioritize during the session.
Suggested flow (if you’re unsure where to start):
🟢 Zone 1 → Zone 2 → Zone 3
The zones progress from more concrete to more conceptual:
- Zone 1 is the most approachable
- Zone 2 introduces more reasoning and design decisions
- Zone 3 focuses on system-level thinking
🟢 Zone 1: Build a Retrieval Agent (Copilot Chat)
Focus on grounding agents in data and producing clear, reliable responses.
🟡 Zone 2: Build a Reasoning & Routing Agent (Copilot Studio)
Design how an agent interprets intent and routes conversations.
🔵 Zone 3: Design & Orchestrate an Agent System(Azure AI Foundry)
Step back and think about agents, workflows, and data as a system.
Each zone includes clear goals, optional guidance, and checkpoints to help you decide when to move on.
- You’ll see suggested prompts and examples throughout — use them as-is or adapt them.
- You’re encouraged to compare approaches with others at your table.
- Coaches will rotate to help answer questions, unblock progress, and discuss design tradeoffs.
- You don’t need to “finish” a zone to move on — partial progress is expected and fine.
Depending on which zones you explore, you may leave with:
- Hands-on experience building and testing agents
- A clearer understanding of intent-based routing and reasoning
- Exposure to agent orchestration patterns in Azure AI Foundry
- New perspectives on how agent systems could support real scenarios
There’s no required output and no single path through the experience.
Use the time to explore what’s most interesting or relevant to you.