-
-
Interfaz Principal
Our Inspiration We were tired of the repetitive cycle of opening Google Sheets, applying filters, and sorting columns just to find a single piece of information. This manual process felt inefficient and broke our workflow. Our inspiration was simple: What if we could just ask our spreadsheets for the data we need? We set out to build a bridge between natural human language and raw spreadsheet data.
How We Built It We developed MCP Sheets Server, a lightweight Node.js application that connects the Gemini AI with the Google Sheets API. The server exposes a set of reliable functions (tools) like get_top_products and get_sales_by_date.
When a user asks a question like, "¿Cuáles fueron las ventas de la semana pasada?", Gemini interprets the request, selects the appropriate tool from our server, and executes it to retrieve the precise information from the sheet.
Challenges We Faced Our main challenges were setting up the secure OAuth 2.0 authentication with Google's API and fine-tuning the prompts to ensure Gemini could reliably choose the correct function and parameters for each user query.
What We Learned This project was a deep dive into creating practical, tool-based AI applications. We learned that the true power of an LLM isn't just generating text, but acting as an intelligent reasoning engine to use other, more deterministic tools. It solidified our belief that conversational interfaces are the future for making data accessible to everyone.
Log in or sign up for Devpost to join the conversation.