Chapter 7.1 - SQLite Chat History

[!info] Tracking chat history persistently.

A RAG architecture is useless if it cannot remember what the user just said. We use a local SQLite database to store user sessions.

import sqlite3

def init_db():
    conn = sqlite3.connect('rag_history.db')
    c = conn.cursor()
    c.execute('''
        CREATE TABLE IF NOT EXISTS messages (
            id INTEGER PRIMARY KEY AUTOINCREMENT,
            user_id INTEGER,
            role TEXT,
            content TEXT,
            timestamp DATETIME DEFAULT CURRENT_TIMESTAMP
        )
    ''')
    conn.commit()

When generating an answer, we pull the last N messages from SQLite and prepend them to the LLM prompt, giving it full conversational context!