codexCEL

The Landing Page
The AI Excel Analyzer

💡 The Inspiration

Small businesses, non-technical managers, and startup founders generate massive amounts of critical data daily inside Excel spreadsheets. However, turning rows of data into actionable trends usually requires complex formulas, tedious pivot tables, or expensive enterprise tools like PowerBI. We built codexCEL to democratize business intelligence—giving anyone the power to talk to their data naturally and see it visualized instantly.

⚙️ How We Built It

We engineered a high-performance, friction-free web interface utilizing a modern, decoupled architecture:

Frontend: A blazing-fast web app built with React and Vite, styled with a crisp white and Microsoft Excel-green theme using custom Tailwind UI primitives to match corporate data environments perfectly.
Backend: A robust Node.js + Express server handling secure file intake.
Database & Parsing: Incoming .xlsx files are parsed into structured JSON objects on the fly using xlsx and tracked locally inside a streamlined SQLite database. This allows us to maintain localized states and pre-generated mock data seamlessly.
AI Integration: We leveraged the OpenAI API (gpt-4o) with specialized system prompts. The model is trained to interpret the dataset and return a dual-payload response: contextual English insights and a structured JSON block mapping chart coordinates for dynamic client-side rendering.

🧠 Challenges We Faced & Lessons Learned

Managing token footprints when feeding large, multi-row Excel sheets into LLM context windows was a core challenge. We overcame this by optimizing our backend extraction layers to transform spreadsheet arrays into incredibly concise, flat JSON formats before routing them to OpenAI.

We also learned how to master deterministic JSON outputs from Generative AI to ensure that our charting libraries (Recharts) can perfectly parse coordinate data dynamically without crashing the frontend layout.

🚀 What's Next for codexCEL

We aim to transition from basic local SQLite files to a fully distributed system on Azure, integrating live multi-agent background workflows that can watch corporate email inboxes, auto-extract incoming attachments, and update business dashboards entirely on autopilot.