Learn how AI agents really work by building them from scratch - from tiny agent to sophisticated deep research agents.

This hands-on workshop demystifies AI agents by teaching you to build them yourself. You'll understand:

• How agents think: The inner workings and architecture of AI agents
• Why prompts matter: Crafting effective prompts for agent behavior
• Tool integration: Building tools that let agents interact with the real world
• Web deployment: Running agents directly in browsers for privacy and offline use

Build your first AI agent with real-world capabilities:

• Learn the core agent loop: think → act → observe
• Integrate tools for real world operations
• Understand prompt engineering for agent behavior

Create a team of specialized agents working together:

• Design agent coordination patterns
• Build inter-agent communication
• Generate comprehensive research reports

Deploy agents that run 100% in the browser:

• Use WebLLM for local AI inference
• Build browser-specific tools
• Create offline-capable AI applications

• ✅ Basic JavaScript (functions, async/await, JSON)
• ✅ HTML/CSS fundamentals
• ✅ Curiosity about AI agents!

• Browser: Chrome/Edge 113+ (for WebGPU support)
• Node.js: Version 18 or higher
• Code Editor: Any (VS Code recommended)
• Groq API Key: Free at console.groq.com

1. Get Your Groq API Key

   • Sign up at https://console.groq.com
   • Navigate to API Keys → Create New Key
   • Save it securely - you'll need it during the workshop

2. Verify Browser Compatibility

   • Check WebGPU support by visiting chrome://gpu or edge://gpu
   • Look for "WebGPU: Hardware accelerated"
   • Test at https://chat.webllm.ai/ (optional)

Note: I will add code to this repo to test Groq API Key and pre-download WebLLM model so that we are prepared during the workshop.

By the end of this workshop, you'll understand:

1. Agent Architecture

   • How agents process information and make decisions
   • The role of prompts in shaping agent behavior
   • Patterns for building reusable agent frameworks

2. Tool Development

   • Creating tools that extend agent capabilities
   • Defining tool schemas and handling responses
   • Best practices for tool error handling

3. Deployment Strategies

   • When to use cloud vs. local inference
   • Privacy considerations in AI applications
   • Performance optimization techniques

• ✅ Complete source code for all three agent types
• ✅ Example tools and integrations
• ✅ Best practices documentation
• ✅ Post-workshop resources and community access

• During Workshop: Live Q&A and hands-on help
• After Workshop: Access to code repository and community
• Questions?: Open an issue in this repo

• WebLLM Documentation
• Tool Calling Best Practices
• Agent Architecture Patterns

This workshop transforms you from JavaScript developer to AI agent architect. No prior AI experience needed - just bring your curiosity and we'll handle the rest!

See you at the workshop! 🚀