← all repositories

langchain-ai/agents-from-scratch

A hands-on guide to building AI agents from scratch using LangGraph, culminating in an email assistant with human-in-the-loop and memory capabilities.

1.8k stars Jupyter Notebook Agents
agents-from-scratch
Velocity · 7d
+4.1
★ / day
Trend
steady
star history

This repository provides a structured tutorial for building AI agents from the ground up using LangGraph. It progresses through four stages: agent basics, agent evaluation, human-in-the-loop patterns, and memory integration. The final project deploys an email assistant connected to the Gmail API, with principles applicable across a wide range of agent tasks. The project uses LangSmith for observability and OpenAI for language model inference.

heatdrop uses Google Analytics to see which pages get read — nothing else. Your call. How we handle data.