Company Research Agent

Inspiration

Manual company research is often tedious, inconsistent, and time-consuming. We were inspired to automate this process using the power of AI and recent advancements in multi-agent systems and large language models like Gemini and GPT-4. Our goal was to build a research assistant that feels like a team of analysts working together behind the scenes.

What it does

The Company Research Agent automates end-to-end company research. It extracts core business information, analyzes market trends, gathers financial metrics, and retrieves relevant news. All this is synthesized into a structured, easy-to-read report in real-time, helping users save time and make better decisions.

How we built it

We used a modular, agent-based architecture:

• Research Agents handle company analysis, industry trends, financial data, and news scraping.
• Processing Agents aggregate, filter (using Tavily), summarize (via Gemini 2.0 Flash), and format content (via GPT-4.1-mini).

The backend is built with FastAPI (including WebSocket support), and the frontend is built in React for real-time updates. We offer multiple deployment options, including Docker and an auto-setup script.

Challenges we ran into

• Filtering out redundant or low-value content
• Maintaining context in large text summaries
• Orchestrating real-time communication between backend and frontend
• Coordinating tasks between Gemini and GPT-4 models
• Designing the system for future scalability

Accomplishments that we're proud of

• Successfully orchestrated multiple LLMs in a single pipeline
• Delivered real-time streaming updates with WebSockets
• Created a developer-friendly setup with modular components
• Built a system that generates clean, structured company reports automatically

What we learned

• How to design scalable multi-agent pipelines
• Best practices for LLM handoffs and task division
• The importance of content filtering, deduplication, and formatting
• Real-time frontend-backend integration using WebSockets

What's next for Company Research Agent

• Expanding support for more data sources and languages
• Adding more customizable report formats
• Improving summarization accuracy with fine-tuned models
• Building API access for third-party integration

Built With

• and
• docker
• fastapi
• google-gemini-api
• langchain
• mongodb
• openai-api
• optional
• python
• react.js
• tailwind-css
• tavily-api
• vite
• websockets