# langchain-ai/agents-from-scratch

Build an email assistant with human-in-the-loop and memory

Repository: https://github.com/langchain-ai/agents-from-scratch
Canonical: https://ross.abutalabs.com/products/agents-from-scratch
Homepage: https://academy.langchain.com/courses/ambient-agents
Language: Jupyter Notebook
License: MIT
License Family: permissive
Topics: agents, memory
Last push: 2026-08-11T15:37:42+00:00

## Health v2 (maintenance only)
Score: 63/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 97, release rhythm 35, longevity 37
- inputs: {"age_days": 520, "days_push": 22, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2156, forks 423 (observed 2026-08-28T04:06:19.613399+00:00)

## What it is
A tutorial repository and companion course from LangChain Academy that teaches how to build AI agents from scratch using LangGraph, culminating in an email-managing 'ambient' agent connected to the Gmail API. It is organized into four progressive sections covering agent basics, evaluation, human-in-the-loop interaction, and memory, with notebooks and accompanying Python source code.

## Use cases
- learn how to build AI agents with LangGraph from scratch
- build an email assistant agent that triages and manages Gmail
- add human-in-the-loop approval to agent workflows
- implement long-term memory in LLM agents
- evaluate agent performance with LangSmith
- learn agent deployment patterns
- understand the difference between agents and workflows

## When to choose
- you want a structured, hands-on tutorial for learning LangGraph and agent patterns
- you are building an email or other 'ambient' agent with human oversight and memory
- you prefer learning through notebooks paired with production-style source code
- you want free course material from the LangChain team

## When to avoid
- you need a production-ready email automation tool rather than educational material
- you want a framework or library to depend on rather than example code
- you are not willing to use OpenAI and LangSmith API keys
- you need agent features beyond the scope of the tutorial without adapting the code yourself

## Facets
- artifact type: learning-resource
- maturity: active
- function: agent-framework, machine-learning, llm-inference, rag
- domain: artificial-intelligence, large-language-models, education, tutorials
- platform: python, cross-platform
- tags: langgraph, langchain, langsmith, email-assistant, human-in-the-loop, agent-memory, jupyter-notebooks, ambient-agents, course-material, ai-agents, automation

## Member repositories
- langchain-ai/agents-from-scratch (main) score 63

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:19.613399+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:50:28.452511+00:00, confidence not recorded.
  - readme: https://github.com/langchain-ai/agents-from-scratch (fetched 2026-08-28T04:06:19.613399+00:00, sha 46672b55e21c)
  - homepage: https://academy.langchain.com/courses/ambient-agents (fetched 2026-08-29T10:30:36.191294+00:00, sha 127bd2897a44)
  - site_page: https://docs.langchain.com/ (fetched 2026-08-29T10:30:36.193699+00:00, sha e6cbe67a291c)
  - site_page: https://www.langchain.com/about (fetched 2026-08-29T10:30:36.203385+00:00, sha b85a33f40436)
  - site_page: https://changelog.langchain.com/ (fetched 2026-08-29T10:30:36.195726+00:00, sha 3c5994039295)
- Data as of 2026-08-30T08:39:29.467469+00:00.