/ / project
An AI stock research agent that turns natural-language investment questions into structured, actionable insights using NSE-linked market data, news retrieval, and agent-based orchestration.
01 / problem
The core challenge in this project was translating ordinary investor questions into an AI workflow that could resolve the right NSE company, fetch the right market data, and summarize meaningful signals from recent news.
Instead of forcing a user to know exact ticker names or navigate multiple sources manually, the agent handles the intent mapping and retrieval process automatically.
02 / approach
03 / architecture
The implementation is organized around a LangGraph StateGraph, which makes the workflow transparent and reusable. Each stage is responsible for a specific step in the pipeline:
04 / results
The project demonstrates how an agent can move from raw prompt input to a useful financial research response. It also includes state persistence using SqliteSaver, which makes replay and debugging much easier during iterative development.
In practice, this makes the agent feel like a research assistant rather than a simple chatbot.
