Multiple Tools in a LangGraph Workflow

Imagine our dinner party again. Last time, we had only one specialized guest, the math expert, ready to jump in when a calculation was needed. But what if you invited a whole squad of specialists? A weather forecaster, a translator, and a trivia buff all joined the party. Now the conversation can flow in richer directions, with the AI router deciding which guest to call depending on the request.

This is exactly what we are building in this lesson: a LangGraph workflow where the AI can choose among multiple tools, not just one.

In real-world applications, a single skill is not enough. Assistants often need to juggle multiple responsibilities:

• Customer support assistant: Needs a knowledge base lookup tool, a ticket logging tool, and possibly a status update tool.

• Research assistant: Should calculate numbers, translate languages, fetch quick facts, and still hold a coherent conversation.

• Productivity bot: Could manage calendars, draft reminders, and pull in external data.

If we packed all of these abilities into one node, the workflow would quickly become messy. LangGraph keeps it clean by separating responsibilities: the AI decides when a tool is needed, tools_condition directs traffic, and ToolNode executes the right helper.

How do we add multiple tools to a LangGraph graph?

We already know how to add a single tool ...