Case study · Own product

Loa — AI event planning

An AI event planner for celebrations in Colombia, in Spanish and English, built end to end with agentic development.

The product

People describe the event they imagine. Loa's planner asks one question at a time — guests, budget, atmosphere, priorities — then shows a preview of the event and, when they're ready, sells a complete professional plan.

It runs on Android and iOS, in Spanish and English, with account deletion, encrypted plan documents and no ads or tracking.

How it was built

A .NET 10 backend in Clean Architecture and a React Native (Expo) app, deployed to Azure. AI goes through ports the product owns, so the planning engine and model providers can change without touching the application layer.

Every piece of work is an increment with numbered sub-increments, a status tracker and a checklist. Each sub-increment leaves the repo green — it builds and every test passes — so work resumes cleanly across sessions, agents and models. Agent rules live in the repo next to the code.

What it shows

A small team with coding agents can carry a real product through payments, store review, two languages and production operations, as long as the plan, the tests and the reviews are not optional.

That is the method we bring to client work.

Back to all work

Contact

Let's build with agents.

Tell us what you're building or what your team is trying. You'll get a straight answer on whether agents fit.