Functions, Tools and Agents with LangChain
by DeepLearning.AI × LangChain
Learn the tool-calling primitive every agent framework is built on, from LangChain's own CEO.
Overview
Taught by LangChain co-founder and CEO Harrison Chase, this DeepLearning.AI short course runs 1 hour 54 minutes across eight video lessons, six of which pair the lecture with a runnable Jupyter notebook, plus a graded quiz. The syllabus moves in a deliberate order: Introduction (2m), OpenAI Function Calling (13m), LangChain Expression Language or LCEL (16m), OpenAI Function Calling in LangChain (12m), Tagging and Extraction (24m), Tools and Routing (17m), Conversational Agent (16m) and Conclusion (1m). The through-line is the tool-calling primitive that every agent framework sits on top of: you see the raw OpenAI function-calling JSON first, then Pydantic schemas for validated structured output, then LCEL's pipe composition for chaining, fallbacks and streaming, then routing across multiple bound tools, and finally a conversational agent that keeps scratchpad state between turns. Because it teaches the primitive rather than one framework's high-level agent constructor, most of it transfers to whatever harness you end up using, and Chase founded LangChain so the LCEL material is first-hand. Two caveats are worth knowing before you start: the mirrored Coursera listing rates 4.7 out of 5 across 126 reviews with learners noting the advertised hour covers video only and the notebooks take considerably longer, and the material is pinned to OpenAI's function-calling API and mid-generation LangChain idioms rather than the newer LangChain v1 agent APIs.
At a Glance
- Topic
- Agentic
- Level
- Intermediate
- Format
- Course
- Cost
- Free
- Duration
- 1 hr 54 min of video, self-paced (reviewers report the notebooks take longer)
- Provider
- DeepLearning.AI × LangChain
- Hands-on
- Yes — code/exercises
- Certificate
- None
What You’ll Learn
- ✓How OpenAI function calling emits structured JSON your runtime can execute
- ✓Defining and validating tool arguments with Pydantic schemas for reliability
- ✓Composing chains with LangChain Expression Language pipes, fallbacks and streaming
- ✓Tagging and extracting structured records from unstructured text using functions
- ✓Routing a single request across several bound tools and handling the results
- ✓Building a conversational agent that keeps scratchpad state across turns
Highlights
- •Taught by Harrison Chase, LangChain's co-founder and CEO — first-hand on LCEL
- •Teaches the tool-calling primitive itself, so the knowledge outlives any one framework
- •Six of the eight lessons ship a runnable notebook alongside the video
- •Under two hours of video, closed out by a graded quiz
- •Free on DeepLearning.AI; the mirrored Coursera listing rates 4.7/5 across 126 reviews
Who It’s For
Best For
- ✓Developers whose agent breaks on unreliable or malformed tool calls
- ✓Engineers who want to understand function calling below the framework abstraction
- ✓Python developers moving from prompt-only LLM apps to tool-using ones
Prerequisites
- •Intermediate Python, including classes and decorators
- •Familiarity with writing prompts against an LLM API
- •An OpenAI API key to run the notebooks outside the course environment
FAQ
What is Functions, Tools and Agents with LangChain?
A 1-hour-54-minute DeepLearning.AI short course from LangChain co-founder Harrison Chase on the primitive every agent depends on: structured function and tool calling. You work through raw OpenAI function calling, Pydantic-validated extraction, LangChain Expression Language composition, and tool routing, then finish by building a conversational agent that selects and invokes its own tools across turns.
Is Functions, Tools and Agents with LangChain free?
Functions, Tools and Agents with LangChain is free to access.
What level is Functions, Tools and Agents with LangChain for?
Functions, Tools and Agents with LangChain is aimed at a intermediate audience. Recommended background: Intermediate Python, including classes and decorators, Familiarity with writing prompts against an LLM API, An OpenAI API key to run the notebooks outside the course environment.
How long does Functions, Tools and Agents with LangChain take?
Expect roughly 1 hr 54 min of video, self-paced (reviewers report the notebooks take longer). Most learners work through it at their own pace.
What will I learn from Functions, Tools and Agents with LangChain?
You'll learn: How OpenAI function calling emits structured JSON your runtime can execute; Defining and validating tool arguments with Pydantic schemas for reliability; Composing chains with LangChain Expression Language pipes, fallbacks and streaming; Tagging and extracting structured records from unstructured text using functions; Routing a single request across several bound tools and handling the results; Building a conversational agent that keeps scratchpad state across turns.
Topics
Sources
This page was written from 2 sources, 1 on domains other than deeplearning.ai.