Multi-Agent System for AI-Powered Market Research, Use Case Generation, and Resource Collection

This project implements a multi-agent system to automate market research, AI/ML use case generation, and resource collection. It uses Langchain, LangGraph, Gradio and Tavily API to generate actionable AI solutions tailored to industries or companies.

Features

• Market Research: Analyze industries or companies, providing insights into trends, competitors, and technology adoption.
• Use Case Generation: Create AI, ML, and automation use cases based on research findings.
• Resource Collection: Gather datasets, tools, and APIs for implementing AI solutions.
• Gradio Interface: User-friendly interface for input, progress tracking, and output retrieval.

Technologies Used

• Langchain: Framework for language model-powered applications.
• LangGraph: Graph-based tool for modeling agent workflows.
• Tavily API: Web scraping for research data.
• Gradio: For building the interactive UI.

System Workflow

The diagram below illustrates the flow of actions taken by the multi-agent system.

Input Query:

• Users enter a query into the Gradio interface.
  • Example: "Analyze the e-commerce industry."

Step 1: Market Research:

• The research_agent:
  • Gathers industry/company insights using GPT-4 and the TavilySearchResults tool.
  • Synthesizes data into a structured output including:
    • Market trends
    • Competitor analysis
    • Quantitative data (e.g., market size, growth rates)

Step 2: AI Use Case Generation:

• The use_case_agent:
  • Takes the research findings as input.
  • Generates five or more structured use cases, including:
    • Use case objectives
    • AI/ML/automation applications
    • Cross-functional benefits

Step 3: Resource Collection:

• The resource_agent:
  • Maps use cases to actionable resources.
  • Outputs a detailed list of:
    • Relevant datasets
    • APIs, tools, and frameworks
    • Generative AI solutions (e.g., semantic search tools, chatbots)

File Output:

• Resource outputs are saved as a timestamped Markdown file in the output directory for easy access and sharing.