for chunk in split_into_chunks(document, size=500):
    embedding = ollama.embed(model="nomic-embed-text", input=chunk)
    cur.execute(
        "INSERT INTO documents (content, embedding) VALUES (%s, %s)",
        (chunk, embedding)
    )

query_embedding = ollama.embed(model="nomic-embed-text", input=query)
cur.execute(
    "SELECT content FROM documents ORDER BY embedding <=> %s LIMIT 5",
    (query_embedding,)
)
context = "\n".join(row[0] for row in cur.fetchall())
response = ollama.chat(model="llama3", messages=[
    {"role": "system", "content": context},
    {"role": "user", "content": query},
])