Raven is an AI‑powered web app and reading companion for avid nonfiction readers who annotate heavily in the margins. It digitized their book highlights and handwritten notes from mobile photos, organized them with AI-generated tags and summaries, and turned their scattered marginalia into a searchable personal knowledge base.
UI / UX
AI
STARTUP
Problem
Two-thirds of tested users said they take notes in the margins of their books. Which means Raven is asking a deeply habitual, analogue behavior to migrate into a structured digital system. The app's core loop: scan a passage, add a highlight, save a note. The loop broke down at the final step: users couldn't reliably find where their notes went after saving, and several couldn't complete the task at all without guessing.
Solution
Raven was targeting serious readers, but its most prominent AI feature is better suited to students reading for exams than readers building long-term knowledge. Both issues were adjacent enough to address together through a revised information architecture for the notes view, prioritizing retrieval and organization over AI-generated summaries. I recommended repositioning the AI summary as an opt-in secondary action rather than the default outcome of saving a note,
The findings pointed to two separate breakdowns: 1. the save confirmation problem. 2. feature-audience mismatch. • 66% of tested users reported taking notes in physical marginalia, establishing a strong prior behavior that the app's digital flow needed to respect rather than replace.