2026 · 3 months · Personal project
Product Designer, UX Researcher, Brand Designer, AI Design Technologist
Beach decision app for the Canary Islands, live on the App Store and Google Play. Every swimming spot - beaches, natural pools, and charcos - gets a real-time 0-100 score on one map, so deciding which beach to go to today, or tomorrow afternoon, takes about 10 seconds instead of five weather tabs.
Islanders lose beach days to the wrong coast: you arrive at a grey, wind-blasted beach while another spot fifteen minutes away is in full sun. Weather apps average by municipality, so they cannot tell those two places apart.
The answer had to arrive in about ten seconds, on a phone, on the way out the door. That rules out a dashboard: three official data feeds had to compress into one number a person trusts without reading a forecast.
A decision layer on top of public weather data: the work is not the pipeline, it is compressing three official sources into one 0-100 score per spot, refreshed every 15 minutes. The sources are Open-Meteo, AEMET (beach forecasts, tides, water temperature, warnings) and NÁYADE (official bathing water quality). Weakest-link logic across a time window: one bad hour drags the score down, because a plan is only as good as its worst hour. Sport modes reshape the whole map, each with its own physics. React Native and Expo in a Turborepo monorepo shared with the Next.js landing.
Solo. Research, scoring model, architecture, app, brand identity, landing, SEO, and both store releases.
Live on the App Store and Google Play, covering 221 swimming spots across the 8 Canary Islands, plus 21 surf breaks and 11 dive sites. Roughly a third of installs are still active users, with no marketing spend, no account and no ads.
Growing up in the Canaries, picking the right beach was instinct: which coast hides from the wind, when the calima rolls in. PlaYa! encodes that local instinct so anyone gets it, not just locals.
The hardest part was making data feel like intuition. The score had to match what a local would tell you, not just be correct, so the result reads like advice rather than a forecast.
Building it solo in three months was only possible with AI-assisted development, but the bottleneck was never the code. It was always product decisions: what to include, how to present it, and when to stop adding and ship.