🤖 SQL AI Agent | Consumer Goods

Engineer user interface equipped SQL AI Agent with MySQL Workbench integration to generate complex queries. This agent will empower Business/Data Analysts to answer important business questions and facilitate decision making.

📌 Table of Contents

• Overview
• Tools & Technologies
• Engineering SQL AI Agent
• Business Questions & Key Findings
• Author & Contact

---

Overview

This project aims to design and develop user interface equipped SQL AI Agents with an integration of MySQL Workbench to generate complex SQL statements. This user-friendly and time efficient program will help Business/Data Analysts to answer ad-hoc requests and facilitate decision making.

• Design the entire workflow and Use CrewAI to develop SQL Agent
• Integrate the agent with MySQL workbench
• Develop a low design user interface (UI) using Tkinter

🛠️ Tools & Technologies

Tool/Platform	Purpose
Google AI Studio	API Key
CrewAI	Develop AI Agents
Tkinkter	User Interface
ChatGPT & Copilot	Code Assistance
Power BI	Visualization
Gamma AI	Presentation

🤖 Engineering AI Agents

Defining Workflow

Developing Agents

Step 1: Generate API Key

Step 2: Prepare and Test LLM

from crewai import LLM
llm = LLM(
    model= "gemini/gemini-2.0-flash",
    temperature= 0.1,
    api_key= GEMINI_API_KEY
)
llm.call("What is the full name of APJ Abdul Kalam")

Step 3: Engineer SQL AI Agent

agent = Agent(
    llm= llm,
    role="MySQL Query Generator",
    goal= "Generate MySQL queries based on user input.",
    backstory= "You are an SQL expert who writes efficient MySQL queries based on user requirements."
)

task = Task(
    description= f"""Generate an MySQL query based on the following user input:
    '''{user_input}'''
    """,
    agent= agent,
    expected_output= "A valid MySQL query that completes the task described in the user input"
)

crew = Crew(
    agents= [agent],
    tasks= [task]
)
result = str(crew.kickoff().raw).replace("```sql", "").replace("```", "").strip()
result

Step 4: Connecting SQL Agent with MySQL Workbench

• Used Prompt 1 to generate code and connect MySQL Workbench

Step 5: Developing Interface

• Used Prompt 2 to integrate agent and develop a Tkinter GUI.

📈 Business Questions & Key Findings

Define business question:

• Business question: Leadership wants to know the total number of newly launched products accross segments and their contribution in revenue.
• Input prompt:

Generate a query to compute the increae in unique product counts in each segments in F.Y. 2021 and
how much revenue the new products contributed in whole of each segment.
Output:
segment | unique_products_2020 | unique_products_2021 | new_unique_products | total_revenue_2021 | revenue_by_new_products | pct_of_total_revenue | rest_revenue

Designing presentation using Gamma AI

• Used Prompt 3 to design presentation slide

• Modified and Changed the final formatting with company standard

📝 Author & Contact

Gaurav Patil (Data Analyst)

• 🔗 LinkedIn