Agentic AI Engineering Lab(Sprint #3) - Agents That Choose: Connect vs Delegate

**What if your AI agent could decide for itself when call a tool, or ask another agent? **

Welcome to Sprint #3 of the Agentic AI Engineering Lab, where we level up from building protocols to building decision-making agents.
In this hands-on lab, you’ll create an autonomous agent that:

• speaks the A2A protocol to collaborate with other agents
• registers to an agent hub for discovery
• connects to an MCP server when tools are needed
• intelligently chooses when to delegate vs when to execute

No lectures. Just systems thinking + real building.

***

### What We’ll Do

• 09:30 - Quick round of introductions
• 09:45 - Recap: AI Agents, MCP Protocol and A2A Protocol
• 10:00 - Build your decision-making agent
• 11:30 - Live demos + technical teardown
• 12:15 - Wrap-up + takeaways

***

### What to Expect

• Hands-on building, not passive listening
• Real agent-to-agent experiments
• Collaboration with fellow builders
• Fast ideas, quick tests, lots of learning
• Unexpected behavior (the fun part)

### What Not to Expect

• Long lectures
• Step-by-step tutorials
• Sales pitches
• Perfect code

This is a lab; we learn by building, breaking, and iterating.

***

### ‍ Who Should Join

Curious developers, tinkerers, and engineers, with basic knowledge of AI Agents.

Bring:

• Laptop with dev setup
• Access to an LLM (Claude / OpenAI / Ollama)
• Comfort building small things in any language

***

### Location

The venue will be announced soon. Stayed tuned!