Selected case studies
NuvAI — AI Lead Assistant for Human Agents
Human-first UX for an AI that helps human real-estate agents manage and qualify leads — the disclosure gate, pacing controls, and human-handoff architecture that determine whether people can trust an AI acting on their behalf. Built on Claude + n8n; voice is an optional layer under evaluation, not the point.
Disclosure design · Pacing as respect · Handoff UX · EU AI Act-aware
Compass — Personal Orientation Layer
An ADHD-first PWA that collapses fifteen scattered state surfaces into one two-minute morning glance. Built after an audit found an ecosystem that was individually healthy and structurally unable to notice its own failures — three of five scheduled processes had been silently dead for months.
Shipped & live · Hard 3-item cap · One next move · "Nothing needs you" as a success state
Access is not permission
I gave AI agents the keys to my accounts, and wrote down what they weren't allowed to do before I did. This is that board, made interactive: set the levels yourself and watch the agent accept, hold, or refuse. Set “Send a DM” to green and it still refuses — because that function was never built.
Runs on the real autonomy file · Capability-absence over policy · Includes the flaw I found in my own system
ProteGO
Dashboard hierarchy redesign, gamification legibility, and a 5-step AR camera capture flow for an AI-powered citizen science monitoring platform. Shipped to MVP. Full case study under NDA.
Shipped to MVP · European space-tech accelerator programme · Client under NDA
BaseLine
Explainable anomaly triage for rail maintenance engineers — closing the gap between when a track defect is detectable and when a human can act on it. Designed around the 2022 Adamuz accident. Full interactive prototype built in Figma Make.
Full interactive prototype · Figma Make · Rail safety
Fifteen more projects
Explorations, concepts, personal systems, and training work — curated out of the way, one click away.
Human-science background. AI-native practice.
I study how people actually work — with systems, with each other, with technology — and then design clear, humane interfaces around what I find. My background runs anthropology, embodied facilitation, and audio production into a design and product practice that has been AI-native from the start.
I think in journeys, not screens. I keep ethics and user agency at the front as interfaces become AI-driven. Four languages (EN / ES / IT / PT). Splits time between New York and Barcelona — available remotely from either location, open to US and EU markets.
Full story →Fixed-scope engagements, ready to start
Two ways to begin — both fixed price, fixed timeline, async-friendly, and delivered as a written report you keep. If you're not sure which fits, book the intro call and I'll tell you honestly, including if the answer is neither. And both audits carry a plain guarantee: if the report doesn't give you at least three fixes worth shipping, you pay half.
“The way the use-case scenarios are understood by Josue and the UX solutions (and the underlying frameworks) formulated demonstrate Josue's ability to cut through the problems leveraging his multi-displicinary expertise. The formulation of Compass drew my attention, particularly.”
UX Audit — Critical Flow · $550
Five days. Pick the flow that matters most — onboarding, checkout, your AI copilot — and get a teardown of everywhere it leaks trust, with prioritized, implementable fixes. Chat, voice, and agent surfaces are home turf.
5 days · Written report + walkthrough call · Fixed price
UX Audit — Full Product · $850
The whole journey — heuristic review, friction map, and a ranked fix list your team can execute without me. Same audit, full-product scope: every screen, every state, one week.
1 week · Written report + walkthrough call · Fixed price