Project Story: ComercIA💡Inspiration

Coming from Mexico, we have spent our lives surrounded by local and informal commerce—from buying vegetables at Sunday street markets and having dinner at taco stands to picking up snacks at the corner store. As we grew up, we identified a massive opportunity. The owners of these businesses generate invaluable data every day, yet that information remains trapped in notebooks, disorganized Excel spreadsheets, or point-of-sale (POS) systems used merely as cash registers. Furthermore, in recent years, many merchants have begun combining e-commerce with physical retail; they sell on platforms like Amazon, Shopify, or Mercado Libre while simultaneously operating stalls at bazaars or brick-and-mortar shops. This results in their operational data being scattered across various channels. The vast majority of these individuals lack a background in finance or accounting. We were inspired by the idea of democratizing advanced analytics and artificial intelligence, delivering them in a simple way so that anyone can protect and grow their business. That is how ComercIA was born. ComercIA.🛠

What it doesComercIA

ComercIA is a financial translator that turns a sea of scattered files into real money and actionable daily decisions for business owners with no technical expertise. Users simply drag and drop their sales, inventory, or purchasing reports from their mobile phones. ComercIA automatically cleans up messy data and displays: Actions for Today—three clear priorities to execute immediately (e.g., what to liquidate to quickly recover cash or what to restock before it runs out); Focus Views—straightforward, clean insights into Cash Flow, Inventory, Margins, and Sales; and Contextual Chat—a natural-language chatbot that allows users to query their business (e.g., "Which products are tying up my capital today?") and receive an executive-level, human, direct, and—above all—easy-to-understand response.

⚙️ How we built it

We designed a hybrid architecture comprising two main components to ensure absolute mathematical precision and a seamless user experience:

a Mobile Frontend (Flutter)—an intuitive interface focused entirely on end-user simplicity, enabling the rapid upload of files (Excel, CSV, folders, and ZIPs) directly from a mobile device—

and a Python Analytics Backend (FastAPI)—the decision-making core. We leave no room for AI hallucinations; the backend validates, normalizes, and correlates datasets using unique product codes. Operational truths are calculated on the server in a strictly mathematical and deterministic manner.

Once exact results are obtained, the Large Language Model (LLM) layer acts solely as a narrator, translating the pre-calculated metrics into clear text, friendly language, and actionable suggestions. All persistence, authentication, and storage of source files and snapshots are managed via the Firebase ecosystem (Auth, Firestore, and Cloud Storage).

🛑 Challenges we ran into

Messy input data: Real-world files from merchants are often broken or inconsistent. Developing a robust normalization engine in FastAPI—capable of cleaning up these discrepancies without causing the application to crash—was our biggest data engineering challenge.

The financial viability of the chat feature: Constant natural language queries can make a SaaS subscription model unviable due to the high token costs associated with LLMs. To address this, we designed an optimized Query Builder: instead of sending the AI the entire business history, the chat translates the user's question into a query targeting the specific variable previously calculated by the backend, passing only that specific fragment to the model to generate the final response.

🎉 Accomplishments that we're proud of

Strict separation between mathematics and AI: We kept heavy analysis on the mathematical side using native Python libraries (pandas and NumPy), avoiding reliance on the LLM for calculations.

We created a system with zero financial hallucinations. Full context persistence: We achieved robust integration with Firebase, allowing users to close the app and return days later without losing the thread of their chat, previous analyses, or business history.

True asynchronous architecture: We implemented efficient asynchronous communication between the backend and frontend, ensuring that heavy data processing does not freeze the user experience in the mobile app.

🧠 What we learned

We learned the value of building software with local empathy. We identified a real problem that affects us closely in our daily lives and provided a genuine solution that truly helps people. We succeeded in transforming complex, raw data into strategies for immediate execution that have a direct, positive impact on a merchant's bank account.

🚀 What's next for comercIA

Direct Integration: Eliminate manual file uploads entirely by enabling direct, real-time connections with e-commerce platforms (Mercado Libre, Amazon, Shopify) and the most widely used POS systems in LATAM.

From recommendation to execution: Add direct action buttons to push notifications. If the system detects overstock, ComercIA will automatically draft a promotional campaign ready for the business owner to send to customers via WhatsApp with a single tap.

Specialized AI agents: Add a more robust layer of intelligent agents capable of acting as dedicated business consultants, guiding the merchant through tactical decisions for the medium and long term.