Build a Text-to-SQL Agent Using the SmolAgents Library

I recently started experimenting with the SmolAgents library from Hugging Face.

It is a nice library for building powerful AI agents quickly.

I created a course for it:

Lycee AI

In the course, I go into detail about the core abstractions of the smolagents library and explain everything with practical examples.

In this tutorial, I am going to quickly present how to build a text-to-SQL agent using smolagents.

The setup is simple.

I am going to use a fake SQLite database, feed some employee data, then create an AI agent that runs SQL queries on the database to reply to user queries.

Here is how to set up the database:

import os
import sqlite3
from smolagents import Tool, ToolCallingAgent, LiteLLMModel, GradioUI
from dotenv import load_dotenv

load_dotenv()

# ----- Step 1: Build an Enhanced SQLite Database -----
db_file = "fake_db.sqlite"
conn = sqlite3.connect(db_file)
cursor = conn.cursor()

# Create an enhanced…