Market Intelligence Agent is a Streamlit-based multi-agent system powered by Groq's LLaMA 3.3 70B model, Tavily Search API, and LangGraph.
The application ingests a business-context PDF, performs automated
competitor research, generates a structured SWOT analysis, and compiles
a final intelligence report.
A graph-based orchestration loop dynamically routes execution between
specialized agents (Research, SWOT, Report) using LLM-driven decision
logic.
The system also provides an interactive chat interface that allows users to query insights directly from the generated intelligence report.
This project demonstrates practical implementation of: - Graph-based multi-agent orchestration (LangGraph) - LLM-driven conditional routing - Autonomous web research using Tavily API - Structured SWOT generation - Context-aware conversational Q&A over generated reports
- LangGraph Multi-Agent Architecture
- Orchestrator-driven control flow
- Conditional routing between:
- RESEARCH_AGENT
- SWOT_AGENT
- REPORT_AGENT
- DONE state
- PDF Business Context Ingestion
- Extracts text using
pypdf - Uses document context to guide research and analysis
- Extracts text using
- Autonomous Competitor Research
- Generates optimized search queries via LLM
- Performs advanced web search using Tavily API
- Aggregates structured competitor intelligence
- SWOT Analysis Generation
- Produces structured Markdown SWOT table
- Integrates business context and competitor data
- Final Intelligence Report Compilation
- Consolidates SWOT + research findings
- Generates structured, readable market intelligence output
- Context-Aware Chat Interface
- Enables interactive Q&A over generated report
- Uses full research + SWOT context for responses
- Iteration Control & Safety
- Maximum research iteration handling
- Robust JSON parsing for LLM decisions
- Fallback termination logic
- Execution Logging
- Step-by-step agent reasoning logs
- Visible audit trail in UI