Breathe: Built with AI | Estefan Saucedo

01 — The challenge

Most breathing apps overwhelm new users before they ever take a breath.

Onboarding flows, account walls, and dense technique menus create friction at exactly the moment someone is reaching for calm. I wanted to test a hypothesis: that thoughtful content design, not more features, is what makes a wellness tool usable. The problems I set out to solve:

Choice paralysis. Mindfulness apps offer an overwhelming amount of choices. I wanted this app to focus on breathing while keeping it very simple and free.
Unclear value. Help users understand which technique fits their needs, depending on their situation.
Friction to first breath. Every tap between opening an app and starting a session is a barrier I wanted to lower.

02 — Process

Research, not a guess at better labels.

Working solo without a research budget, I grounded decisions in established breathing science and content design principles, then pressure-tested the copy against real emotional scenarios: "I'm anxious right now, which one do I pick?" "I'm lying in bed trying to fall asleep." "I need to focus before a meeting." For each, I asked whether the labels and descriptions answered the user's real question — what will this do for me?

I used Claude Code as a hands-on development partner, which let me move from concept to a working, deployed app as a one-person team. The collaboration was deliberate, not hands-off: I directed the architecture and made every content and UX decision, while using AI to speed up the build, debug, and iterate on the front end. This mirrors how I think AI should work in practice — supporting human judgment, with a human owning the calls that matter. It's the same approach I bring to content systems at scale: AI handles speed, the designer owns intent and quality.

Building the Breathe app at a desk, working alongside Claude Code — Working solo, AI as development partner

03 — Solution

Content design and IA. Three principles shaped the final experience.

Lead with benefit, not jargon. Each technique pairs its clinical name with a plain-language outcome — for example "Calms anxiety and improves focus" and "Perfect for falling asleep." The name builds credibility, the benefit line drives the choice.
Match content to emotional state. The four techniques map to four common needs (calm, sleep, stress relief, energy), so users pick based on how they feel rather than needing to know breathing terminology.
Reduce everything between intent and action. No accounts, no onboarding, no settings maze. Pick a technique, hit Start, breathe. The interface gets out of the way.

Before and after comparison of technique cards — jargon-led names versus benefit-led descriptions — Before / after — benefit-led technique cards

04 — Impact

Making a difference with my app.

This project shows two things I care about as a content designer: that content design can carry an entire product experience, and that AI lets one person own that experience end to end without sacrificing craft.

By turning clinical techniques into benefit-led language and removing every barrier to the first breath, I built a tool that works in the exact moment of stress it's designed for.

The best sign came on its own: a therapist told me one of her clients now uses the app every night to wind down for sleep. For a tool meant to help people in hard moments, someone using it as part of their nightly routine is the proof that matters most.

Text message from a therapist saying a client in distress found the app helpful and now uses it as part of their nighttime routine — Unprompted, from a therapist using the app with clients

Use the app for free →