Sun Pharma: AI-Driven Data Insights with OpenAI Autogen

Business Problem

  • Difficulty in retrieving insights from relational databases (RDBMS) containing sales, stock levels, and agent performance data. 
  • Dependency on manual processes for sales analysis and profit evaluation, making decision-making slow and effort intensive. 
  • Lack of a natural language interface for business users to query data intuitively. 
  • No real-time automated chart generation for quick data visualization. 
  • Absence of a dedicated graph tool for customized data representation without building a full dashboard. 

About Atria

Sun Pharma, a global leader in pharmaceuticals, faced challenges in analyzing and extracting insights from its vast sales data. To overcome this, they sought an AI-driven solution that could provide real-time insights, automate data visualization, and enhance decision-making.

Explore Now

Solution

To address these challenges, we developed a Proof of Concept (PoC) leveraging OpenAI and Autogen’s multi-agent framework, PostgreSQL as the database, and Streamlit for visualization. This system enabled intelligent collaboration between specialized AI agents to process and respond to user queries efficiently. 

1.

Conversational Agent – Built using OpenAI and Autogen, this chatbot interprets natural language queries and interacts with other agents. 

4.

Visualization Agent – Uses Streamlit to generate automated charts, making data insights more accessible. 

2.

Query Agent – Dynamically converts user queries into SQL statements to retrieve data from PostgreSQL. 

5.

Graph Tool Agent – A lightweight, Power BI-inspired module integrated with Streamlit to generate custom graphs without requiring a full dashboard. 

3.

Analysis Agent – Processes and structures the retrieved data into meaningful insights, enhancing decision-making. 

6.

LLM Used: The solution leveraged OpenAI’s GPT-4 within the Autogen framework to power natural language understanding, query generation, and data analysis. This enabled seamless multi-agent collaboration for accurate and efficient insights extraction. 

Architecture

  • User Interaction: Business users interact with a chatbot powered by OpenAI and Autogen. 
  • Data Processing: 
    Query Agent dynamically generates SQL queries. 
    Analysis Agent refines and structures the fetched data. 
  • Visualization: 
    Visualization Agent creates automated charts using Streamlit. 
    Graph Tool Agent provides an intuitive interface for graph generation. 
  • Backend Processing: OpenAI and Autogen orchestrates interactions between the agents for seamless data flow and optimized performance. 
Outcomes

0%

Faster Data Retrieval: Business users can access complex insights instantly without requiring SQL expertise. 

0%

Reduction in Manual Effort: Automating sales analysis and visualization significantly cuts down manual intervention. 

0%

Simplified Data Representation: The dedicated graph tool provides a lightweight alternative to traditional BI dashboards.