Rivva

Context

Rivva is an AI scheduling platform that uses biometric data from Apple Health to build energy-aware daily plans — not just calendars, but schedules informed by how the user actually feels. The product was in prototype when the founding team brought in design: a functional ML model existed, but no interaction layer, no onboarding, and no way for a user to understand or trust what the AI was suggesting.

Problem

The ML model could generate a schedule. The design problem was that users had no reason to follow it. Biometric data creates a power asymmetry — the system knows something the user doesn't — and the first instinct in most AI tools is to hide that asymmetry rather than explain it. Early concept testing confirmed users would accept AI suggestions they could understand and reject ones they couldn't, regardless of the quality of the recommendation.

Approach

Designed the full interaction layer from 0→1: Apple Health onboarding, energy-aware scheduling model, and the daily planning experience across iOS, Android, and web. Co-designed Nia, the AI assistant, as an explanation layer — translating ML outputs into language users could act on and disagree with. Built the cross-platform design system that unified the experience across mobile and web. Collaborated with the ML team on how the model's reasoning could be surfaced without overwhelming the interface.

Outcomes

Rivva launched from beta to full release and reached #4 Product of the Day on Product Hunt in its first week. 500+ downloads came in the first month, and 30+ paying customers converted within two months — without a paid marketing budget. The AI explainability layer held: post-launch user feedback consistently cited trust in Nia's recommendations as a reason for staying with the product.