Interactive Notebooks

Run all 25 tutorial notebooks directly in your browser using JupyterLite — no installation required.

Open JupyterLite

Available Notebooks

Phase 1: Core Patterns

Notebook	Description
01_chatbot_basics.ipynb	Build your first LangGraph chatbot
02_tool_calling.ipynb	Implement ReAct pattern with tools
03_memory_persistence.ipynb	Add conversation memory
04_human_in_the_loop.ipynb	Human approval workflows
05_reflection.ipynb	Self-improving agents
06_plan_and_execute.ipynb	Planning and execution patterns
07_research_assistant.ipynb	Multi-step research agent

Phase 2: RAG Patterns

Notebook	Description
08_basic_rag.ipynb	Basic retrieval-augmented generation
09_self_rag.ipynb	Self-correcting RAG
10_crag.ipynb	Corrective RAG with web fallback
11_adaptive_rag.ipynb	Query-adaptive retrieval
12_agentic_rag.ipynb	Agent-driven RAG
13_perplexity_clone.ipynb	Build a Perplexity-style search

Phase 3: Multi-Agent Patterns

Notebook	Description
14_multi_agent_collaboration.ipynb	Agent collaboration basics
15_hierarchical_teams.ipynb	Hierarchical agent teams
16_subgraphs.ipynb	Modular graph composition
17_agent_handoffs.ipynb	Agent-to-agent handoffs
18_agent_swarm.ipynb	Swarm-style agent systems
19_map_reduce_agents.ipynb	Parallel agent processing
20_multi_agent_evaluation.ipynb	Evaluating multi-agent systems

Phase 4: Advanced Reasoning

Notebook	Description
21_plan_and_execute.ipynb	Advanced planning patterns
22_reflection.ipynb	Advanced reflection techniques
23_reflexion.ipynb	Reflexion pattern implementation
24_lats.ipynb	Language Agent Tree Search
25_rewoo.ipynb	ReWOO reasoning pattern

---

How It Works

JupyterLite runs entirely in your browser using WebAssembly:

• No server required — Everything runs client-side
• Instant startup — No waiting for kernel initialization
• Persistent storage — Your work is saved in browser storage
• Full Python — Powered by Pyodide (Python 3.11)

Limitations

JupyterLite cannot make network requests to external services. Notebooks that require Ollama or other API calls will show simulated outputs. For full functionality, run notebooks locally with langgraph-local serve.

Running Locally

For the complete experience with actual LLM calls:

# Clone the repository
git clone https://github.com/AbhinaavRamesh/langgraph-ollama-tutorial
cd langgraph-ollama-tutorial

# Install dependencies
pip install -e ".[dev]"

# Start the local environment
langgraph-local serve

This launches Jupyter with Ollama integration for real LLM interactions.