Before and after: EchoNote shifts to voice-first UX and doubles daily active users

AI · 6 min read

EchoNote's original app centered on typed notes with a secondary voice capture option that required cloud processing and a slow round-trip. Many users reported friction and privacy concerns, and weekly active user growth plateaued. The team hypothesized that a voice-first experience with local transcription would lower friction and appeal to hands-free scenarios.

Designers simplified the home screen to a single big record CTA, added a quick-action widget for lock-screen recording, and created a live transcription overlay that appears during recording with inline editing. Privacy was baked into the UX: by default, transcription runs on-device using a compact ASR model, and sharing or cloud sync is opt-in with clear consent messaging.

Engineering invested in model optimization and selective cloud fallback for edge cases like multi-speaker transcripts. The product also introduced a semantic search using lightweight embeddings computed on-device and periodically synced encrypted deltas to the cloud for advanced queries. Moderated tests and accessibility audits confirmed the new flow reduced motor demand and improved discoverability for screen-reader users.

After a staged rollout, DAU rose 102 percent and retention at 14 days increased 38 percent. Privacy-respecting features were a competitive differentiator during enterprise conversations. EchoNote now treats voice-first affordances as core product primitives and continues to iterate on live editing and multi-language support.