I will have 20,000 fields of data in MySQL to start.. When GPT?

Hello all and thanks in advance for your reply.

I have a fairly complex project that I’m working that has a decent amount of data in the beginning and will grow pretty heavily over time. I assume putting data into a database for ease of access and management is the way to go. And if so, when will I be able to have GPT4 access this data through the API?

Thanks!_TP :sunglasses:


I’m already accessing my data (mysql) asking questions in natural language.
Works quite fine, even if, sometimes the prompt exceed the token limits (i’m using gpt-4 api).
Don’t really know if it fits your needs !?


Hi what you describe it´s exactly what i need, could you share some light on your aproach please?
Thanks in advance

@s.flandrin How are you doing this? Do you pass the schema as context?

import os
import tkinter as tk
import tkinter.ttk as ttk

from langchain.sql_database import SQLDatabase
from langchain.llms.openai import OpenAI
from langchain import SQLDatabaseChain
from langchain.prompts.prompt import PromptTemplate
from langchain.chains import SQLDatabaseSequentialChain

from langchain.chat_models import ChatOpenAI
from langchain.prompts.chat import (
from langchain.schema import (

#Define llm variables

os.environ[‘OPENAI_API_KEY’] = “Your API Key”
llm = ChatOpenAI(model_name=“gpt-4”, model=“gpt-4”, temperature=0)

#Connect to my local database

db = SQLDatabase.from_uri(“mysql://login:Password@”)

#Prompt template

_DEFAULT_TEMPLATE = “”" Let’s think step by step.
Given an input question, first create a syntactically correct {dialect} query to run, then look at the results of the query and return the answer.

Use the following format:

Question: “Question here”
SQLQuery: “SQL Query to run”
SQLResult: “Result of the SQLQuery”
Answer: “Final answer here”

Only use the following tables:


Question: {input}“”"

PROMPT = PromptTemplate(
input_variables=[“input”, “table_info”, “dialect”], template=_DEFAULT_TEMPLATE

#Create the chain with the database and the prompt

db_chain = SQLDatabaseSequentialChain.from_llm(llm, db, verbose=True)

Create the UI window

root = tk.Tk()
root.title(“Chat with your Tabular Data”)

Create the text entry widget

entry = ttk.Entry(root, font=(“Arial”, 14))
entry.pack(padx=20, pady=20, fill=tk.X)

Create the button callback

def on_click():
# Get the query text from the entry widget
query = entry.get()

# Run the query using the agent executor
result = db_chain.run(query)

# Display the result in the text widget
text.delete("1.0", tk.END)
text.insert(tk.END, result)

Create the button widget

button = ttk.Button(root, text=“Chat”, command= on_click)
button.pack(padx=20, pady=20)

Create the text widget to display the result

text = tk.Text(root, height=10, width=60, font=(“Arial”, 14))
text.pack(padx=20, pady=20)

Start the UI event loop


@venk : No I don’t. I am working on a “Description table” of my database (which contains quite a lot of tables and a lot of datas)
Then I’ll try to pass it to the context in order to check if it resolves my problem