AI SQL Assistant using Python & Llama 3

Step 1 - Install Ollama

Download Ollama:

https://ollama.com/download

Verify installation:

ollama --version

Step 2 - Download a Model

ollama pull llama3.2:1b

Start Ollama if it isn't already running:

ollama serve

Step 3 - Create Project

mkdir sql_ai_assistant
cd sql_ai_assistant

Create a virtual environment.

Windows

python -m venv venv

venv\Scripts\activate

Mac/Linux

python3 -m venv venv

source venv/bin/activate

Step 4 - Install Packages

pip install -U langchain
pip install -U langchain-community
pip install -U langchain-classic
pip install -U langchain-ollama
pip install -U sqlalchemy

Verify:

pip list

Step 5 - Create the Database

Create create_db.py

import sqlite3

conn = sqlite3.connect("shop.db")
cursor = conn.cursor()

cursor.execute("""
CREATE TABLE IF NOT EXISTS orders(
    id INTEGER PRIMARY KEY AUTOINCREMENT,
    customer_name TEXT,
    amount REAL,
    order_date TEXT
)
""")

orders = [
    ("Alice",120,"2026-06-01"),
    ("Bob",75,"2026-06-02"),
    ("Charlie",250,"2026-06-03"),
    ("David",99,"2026-06-04"),
    ("Emma",180,"2026-06-05")
]

cursor.executemany(
    "INSERT INTO orders(customer_name,amount,order_date) VALUES(?,?,?)",
    orders
)

conn.commit()
conn.close()

print("Database created!")

Run:

python create_db.py

You'll get:

shop.db

Step 6 - Create main.py

from langchain_community.utilities import SQLDatabase
from langchain_ollama import OllamaLLM
from langchain_classic.chains.sql_database.query import create_sql_query_chain
from langchain_community.tools import QuerySQLDatabaseTool

# Connect database
db = SQLDatabase.from_uri("sqlite:///shop.db")

# Load local model
llm = OllamaLLM(model="llama3")

# Create SQL chain
chain = create_sql_query_chain(llm, db)

executor = QuerySQLDatabaseTool(db=db)

user_question = "Show all orders above $100"

prompt = f"""
You are an expert SQLite SQL generator.

Convert the user's question into SQL.

Rules:
- Return ONLY SQL.
- No explanation.
- No markdown.
- No quotes.
- No SQLQuery label.
- Only one SQL statement.

Question:
{user_question}
"""

sql = chain.invoke({"question": prompt})

print("Raw Output:")
print(repr(sql))

# Clean Output
sql = sql.strip()

sql = (
    sql.replace("```sql", "")
       .replace("```", "")
       .replace("SQLQuery:", "")
       .replace("SQL:", "")
       .strip()
)

# Remove surrounding quotes
sql = sql.strip('"').strip("'")

# Keep only SQL
keywords = [
    "SELECT",
    "INSERT",
    "UPDATE",
    "DELETE",
    "CREATE",
    "ALTER",
    "DROP",
    "WITH"
]

upper = sql.upper()

for word in keywords:
    idx = upper.find(word)
    if idx != -1:
        sql = sql[idx:]
        break

sql = sql.rstrip(";") + ";"

print("\nGenerated SQL:")
print(sql)

print("\nResult:")
result = executor.invoke(sql)

print(result)

Step 7 - Run

python main.py

Expected output:

Raw Output:
'SELECT * FROM orders WHERE amount > 100'

Generated SQL:
SELECT * FROM orders WHERE amount > 100;

Result:
[(1, 'Alice', 120.0, '2026-06-01'),
 (3, 'Charlie', 250.0, '2026-06-03'),
 (5, 'Emma', 180.0, '2026-06-05')]