andefi was developed as a solution to various weaknesses I’ve seen repeatedly in overseas financial markets after having traveled extensively and having grown up in several different foreign nations. The process of creating andefi showcases the implementation of AI-accelerated development practices in a very real, practical way.
Using an amalgamation of AI coding assistants like Cursor, and large models like Claude, GPT, and Gemini, I built a platform that offers tools that simply do not exist in today’s Latin American financial landscape.
Tools Born from Necessity
At its core, andefi is a collection of solutions to financial friction points I’ve personally witnessed abroad. Among them are:
- Pagomático
An automated bill payment platform that connects with Colombia’s Puntored payment network and beyond. Pagomático is designed to handle recurring payments with intelligence — pulling data from utility companies, telecom providers, and other services, then automating the process of settling bills across both Colombian and U.S. payment methods. Instead of logging into multiple portals and remembering payment deadlines, Pagomático manages the entire process in one place. - Lúmina
A financial insights and recommendation engine that leverages Open Finance data across Latin America. By analyzing transactions and patterns, Lúmina helps users understand their financial habits, highlights opportunities for savings, and suggests more efficient use of credit and banking products. It’s not just a budgeting tool — it’s a decision-making companion built on top of emerging Open Banking laws in the region.
These concepts have been accelerated by regulatory changes in Latin America, where countries are adopting open banking frameworks at a rapid pace. andefi exists to take advantage of this moment and bring tools to market that simplify, unify, and modernize financial management.
Making Everything Easier
The goal was always to make financial life easier — not only for users but also in how I built the platform itself. My own development process mirrored that same philosophy of acceleration and simplification.
I began with Figma drafts, creating initial layouts that captured the core product experience. From there, I used AI-assisted tools to translate those designs directly into code. Leveraging Radix component libraries, I was able to maintain a consistent, accessible design system while still moving quickly.
Technology Stack
On the technical side, andefi is built on a modern, serverless architecture:
- Frontend: React, with Radix as the component library.
- Backend: Node.js, running on AWS Lambda.
- Database: PostgreSQL with Kysely for type-safe queries.
- Routing: A custom router designed for compatibility with AI processes.
- State Management: TanStack Query for managing server state.
- Validation: Zod for schema validation and safety.
- Authentication & User Management: Supabase.
This stack ensures flexibility, scalability, and speed, without sacrificing maintainability.
Integrations
A platform like andefi cannot exist in isolation. To process payments, pull financial data, and interact with users, we integrated with several third-party providers, including:
- Puntored – Bill payments across Colombia.
- Brevo – Transactional communications and customer engagement.
- Movii – Digital wallet integrations.
- Additional APIs to bridge payment rails between Colombia and the U.S.
These integrations make the platform not just functional, but powerful in its ability to connect disparate financial systems.