• Overview of multi-agent systems and real-world use cases
• Introduction to LangChain and LangGraph
• Workshop objectives and output expectations

• Setting up Google Colab or VS Code
• Installing dependencies (langchain, langgraph, openrouter, serpapi, etc.)
• Securing API keys using .env files

• Define use case: Startup Advisor Bot- Identify subtasks: idea generation, competitor analysis, GTM planning- Build prompts and simple function wrappers
• Discussion: Division of labor between LLM and code logic

• Use SerpAPI or DuckDuckGo search tool
• Structure output in a predictable format (e.g., tables of competitors)
• Emphasize structured data retrieval vs traditional web search

45-minute lunch or refreshment break

• Brief overview of LangGraph (nodes, edges, states) - Define graph flow: start → idea → research → gtm → end
• Add optional conditional path (e.g. pitch deck agent)
• Visualize and execute graph

• Use a simple in-memory or JSON store to retain state
• Implement load/save conversation features
• Session switching, continuation, and threading

• Adding manual approval steps (e.g., "Would you like to refine this idea?")
• Basic CLI prompts or Colab widgets for interaction
• Integration points for feedback loops

• Run the end-to-end Startup Advisor Bot
• Try diverse domains (e.g., climate, SaaS, retail)
• Highlight how modular agents communicate and share state

• Summary of key concepts and capabilities
• Share complete project code, diagrams, and documentation
• Discuss advanced extensions: UI deployment, multi-user systems, evaluation methods

Wrap up with refreshments, connect with peers and mentors and receive your certificate of participation.