At a Glance

Name: Build a Multi-Tool AI Agent Using LangChain and IBM Granite
Price: Free CAD
Rating: 0.0 (0 reviews)
Author: jigisha_barbhaya, joseph_santarcangelo, anna_gutowska, karan_goswami

Agentic AI is very popular. Learn to build a responsive AI agent using LangChain’s tool calling with IBM's Granite model in 30 minutes. This beginner-friendly hands-on project guides you through combining key tools like live weather data fetchers, YouTube content handlers, and search utilities. Learn to design a multi-tool workflow that delivers dynamic, real-time outputs using LangChain and watsonx.ai. Ideal for developers seeking applied experience in orchestrating intelligent agents with LLMs and external APIs.

Imagine building an AI agent that doesn’t just respond with pre-trained knowledge—but thinks, reasons, and takes action using live data. That’s exactly the journey you’re about to embark on in this guided project.

You’ll start by diving into the world of LangChain, a framework designed to empower large language models with tool calling abilities. Paired with the enterprise-grade intelligence of IBM’s Granite models through watsonx.ai, you won’t just be working with static LLMs—you’ll be training your agent to interact with the world.

Your mission? To craft an agent that can:

Check the weather for any city in real-time,
Search YouTube for relevant videos based on user prompts,
And even perform an iconic search, pulling context-rich information about public figures, places, and events.

Each feature you build will feel like teaching your AI how to use a new sense—sight, memory, or intuition. And thanks to LangChain’s ReAct framework (Reasoning + Acting), your agent won’t just call these tools randomly. It will evaluate the problem, decide which tool is best suited, call it, interpret the result, and return a well-reasoned answer. Step by step, you’ll watch your agent evolve into a thoughtful assistant capable of chaining together multiple tools to solve complex queries—almost like a detective gathering clues before solving a case.

A Look at the Project Ahead

After completing this lab you will be able to:

Understand the concept of tool calling (function calling).
Explain the purpose of the LangChain framework
Implement a LangChain agent using Granite-3B-Instruct model.
Integrate and configure prebuilt tools in LangChain

What You'll Need

Let your learners know what technology and skills they'll need prior to starting this guided project. Remember that the IBM Skills Network Labs environment comes with many things pre-installed (e.g. Docker) to save them the hassle of setting everything up. Also note that this platform works best with current versions of Chrome, Edge, Firefox, Internet Explorer or Safari.

Offered By: IBMSkillsNetwork

Build a Multi-Tool AI Agent Using LangChain and IBM Granite

At a Glance

A Look at the Project Ahead

What You'll Need