🚀 Concept

Build FinSense, an autonomous financial research copilot that behaves like a self-directed junior analyst: it understands complex financial questions, decomposes them into tasks, calls the right data tools, synthesizes insights, and justifies its reasoning — all while chatting naturally with the user.

You can think of it as “Perplexity + Bloomberg Terminal + Mini-GPT-Researcher”.

🎯 Objectives

Design an agentic system that:

1. Understands multi-turn queries about stocks, macro indicators, or portfolios.
2. Chooses & orchestrates tools (finance APIs, data scrapers, calculators) dynamically.
3. Synthesizes data into coherent, contextual insights.
4. Explains its reasoning as it works — not just answers, but why.
5. Adapts over time — remembers previous context and adjusts behavior.

🧱 Core Features

1. 🧩 Agent Framework

Implement a reasoning-action loop:

User Input → Planner → Tool Calls → Data Aggregator → Summarizer → Response

• You may use any open source model for reasoning and function calls.
• Implement at least three tools:
  • get_stock_price(symbol, period) – pull data from Yahoo Finance / Alpha Vantage / Finnhub.
  • get_macro_data(metric) – CPI, GDP, Fed Rate, etc. (FRED or mock API).
  • get_finance_news(topic) – from NewsAPI or RSS feeds.

(Bonus tools below.)