LLM Mesh agentic applications

This tutorial series demonstrates how to build agentic applications using the LLM Mesh in Dataiku. Through an example of building a customer information assistant, you’ll explore three parts of an agentic workflow:

  1. Tools - How to define and use tools with the LLM Mesh

  2. Agents - How to create an LLM-based agent that uses multiple tools

  3. Webapps - How to build a web application with the agent

Prerequisites

  • Dataiku >= 13.2

  • LLM Mesh connection to a provider that supports tool calls (tested with OpenAI GPT 5.1; you can find a compatible list here)

  • An SQL Dataset named pro_customers_sql. You can create this Dataset by uploading this CSV file and using a Sync recipe to store the data in an SQL connection.

  • Python >= 3.12

  • A code environment with the following packages

    ddgs        # tested with 9.13.1