Content design and design system for Umbra

Umbra is an early-stage infant sleep support app for parents looking for accurate, non-judgmental answers about normal infant sleep. Most of the category leans on fear, absolutes, and sales pressure. Umbra's whole premise is the opposite: honest, evidence-based guidance delivered at the exact moment a tired parent is looking for it, often at 3 a.m. on a phone in a dark room.

Content has to demystify a complex, emotionally loaded subject without oversimplifying it, carry a parent through a long, high-consideration journey from first search to subscription, and do it in a voice consistent enough to scale across thousands of pieces of guidance.

Ongoing. The product was designed and prototyped in Lovable, with the content system built in parallel so language and interface evolved together rather than copy arriving as a finishing touch.

There was no voice, no vocabulary, and no standards to inherit — everything started at zero. The content also had to hold up under real-world conditions: an exhausted parent up stressing in the middle of the night, ping-ponging through misinformation and hype and false promises, looking for support. That meant designing for a closed-loop chatbot that only answers from a vetted content library, a dark mode built for late-night stress scrolls, and language that emphasizes care and consent over shame.

• Built the content style guide and content design system from scratch: voice and tone principles, terminology decisions, message patterns, error and empty states, and review standards for anyone writing in the product.
• Established a component-level content model so every note, prompt, and notification maps to a reusable pattern rather than being written one screen at a time.
• Designed and prototyped the product experience in Lovable, iterating on language and interface together in a rapid, collaborative loop.
• Built the narrative backbone across the purchase journey — onboarding, daily notes, ask, resources, subscription, and trial messaging — so every surface shares a story.
• Designed for accessibility first: plain language, front-loaded key messages, clear content hierarchy, tap-target and contrast standards, and full parity between light and dark modes.
• Shaped the closed-loop chatbot experience, scoping it to answer only from the vetted content library with clear sourcing and guardrails so guidance stays accurate and never improvises.
• Wrote non-judgmental recommendation frameworks that name what is normal, explain the developmental why, and offer options rather than verdicts.
• Used AI tooling to ideate screen and content concepts and to structure prompts for consistent on-brand drafting, while keeping human judgment and human writing on everything that ships.

Content design as infrastructure: a system, a vocabulary, and a set of standards that let a small team say something honest and useful, thousands of times over, in exactly the same voice.