DealWitIt

About the project

Inspiration

The inspiration for DealWitIt comes from a critical vulnerability in the high-stakes world of mergers and acquisitions: the staggering rate of failure due to unforeseen cultural clashes. Billions of dollars are lost when deals that look perfect on paper crumble under the weight of incompatible brand identities and misaligned audiences. The project was born from the idea of replacing executive "gut feeling" with a data-driven, quantifiable approach to cultural due diligence. The goal was to create a financial-grade intelligence platform that could de-risk acquisitions by analyzing the very human element of a deal.

How It Was Built

DealWitIt is built on a modern, decoupled architecture:

Backend: A powerful FastAPI server acts as the intelligence engine. It handles user authentication (including Google OAuth), orchestrates a complex workflow of API calls, and runs the core of the platform: a sophisticated ReAct (Reasoning and Acting) agent. This agent, powered by Google's Gemini, systematically gathers data Tavily's Search API to build a comprehensive cultural and financial profile of the companies being analyzed. All data is persisted in a PostgreSQL database, managed with SQLModel for asynchronous operations.
Frontend: A sleek and professional Next.js application serves as the executive dashboard. Built with TypeScript, React, and styled with Tailwind CSS and Shadcn UI, it provides a seamless user experience for initiating analyses and exploring the rich, multi-faceted reports. The frontend communicates with the backend via a REST API and uses SWR for real-time data fetching and Framer Motion for a polished, modern feel.

Challenges Faced

One of the primary challenges was designing and implementing the autonomous ReAct agent. It required a carefully crafted workflow to ensure it could strategically call upon the various APIs (Gemini, Tavily) in the correct sequence to gather and synthesize the necessary data for a comprehensive report. This involved creating a robust system of prompts and tool definitions to guide the agent's reasoning process. Another significant challenge was ensuring the data returned from the various APIs was correctly interpreted and integrated into the final report, especially when calculating the all-important Cultural Compatibility Score, which is a weighted average of audience affinity and persona expansion potential:

$$ CulturalCompatibility = (0.6 \times AffinityOverlap) + (0.4 \times PersonaExpansion) $$

This required careful data modeling and validation to ensure the final output was both accurate and reliable.

What I Learned

Building DealWitIt was a deep dive into the power of multi-agent systems and the incredible potential of combining specialized AI services. I learned how to orchestrate a ReAct agent to perform a complex, multi-step analysis that mimics the workflow of a human analyst. Integrating provided a powerful lesson in how cultural data can be transformed into a quantifiable business metric. Finally, building a full-stack application with a separate frontend and backend reinforced the importance of clean API design and robust data handling, especially when dealing with asynchronous data streams for a real-time user experience.

Built with

AI & Data: Google Gemini, Tavily Search API
Backend: FastAPI, Python, SQLModel, PostgreSQL
Frontend: Next.js, React, TypeScript, Tailwind CSS, Shadcn UI, SWR, Framer Motion, CopilotKit AG-UI