Competitor Intelligence Agent
An AI-powered agent that ingests unstructured competitor documents (PDFs), extracts structured insights using LLM + RAG, stores them in a SQL database, and enables natural language querying — built with FastAPI, LangChain, ChromaDB, Gemini API, and Docker.
3
Core AI components
129
Pages processed
Docker
Production-ready
The Challenge
Companies dealing with large volumes of competitor research documents (PDFs, reports) struggle to extract structured insights at scale — manual review is slow, inconsistent, and impossible to query programmatically.
Our Solution
Built a three-component AI agent system: a RAG ingestion pipeline (LangChain + ChromaDB) for semantic document retrieval, an LLM extraction agent (Gemini API + Pydantic) for structured data extraction, and a Natural Language to SQL interface for querying extracted data.
- RAG pipeline: PDF ingestion → text chunking → vector embedding → ChromaDB indexing
- LLM extraction agent: Gemini API reads retrieved context → extracts structured JSON → Pydantic validation
- NL-to-SQL interface: natural language question → Gemini generates SQL → SQLite executes → natural language answer
- FastAPI REST endpoints: /ingest, /query, /rag-query, /competitors with filter support
- Full Docker + docker-compose deployment for production-ready containerization
- SQL injection prevention, SELECT-only enforcement, and hallucination mitigation prompts
Tech Stack
The Result
Delivered a fully containerized AI agent capable of processing 129-page competitor reports, extracting structured competitive intelligence, and answering natural language queries like 'Which games lead downloads in Indonesia?' — all via a clean REST API.
Got a project in mind?
Drop us a message and we'll get back to you within 24 hours.
Tell Us About It