Challenge
Helping Readers Understand, Engage, and Apply What They Read
Reading is easy. Truly understanding a book — and carrying its lessons forward — is hard. Audiobooks and summaries increase access but encourage passive consumption. Readers often finish a book without retaining much, or they struggle to connect insights to real-life action.
Bemis was designed to solve this gap: helping readers engage more deeply, extend their interaction with books, and turn insights into actionable takeaways.
What We’re Tracking
We launched Bemis in beta with the goal of understanding how readers naturally use AI to extend their learning and reflection.
Engagement Depth
Average number of back-and-forth exchanges per book session.
Insight Retention
% of sessions resulting in saved or edited notes.
Action Potential
Frequency of users exporting notes or bookmarking books for later.
Use It Anytime You Need to Reflect
Bemis helps readers shift from passive listening to active engagement. Here’s how users consistently find value:
1. Ask Questions Mid-Read
Instead of passively consuming, readers ask about themes, chapters, or confusing sections — and get conversational insights in real time.
2. Key Insights on Demand
Each discussion produces structured insights, short and shareable, tied directly to the book.
3. Notes That Grow With You
AI-generated notes serve as editable scratchpads, allowing readers to refine thoughts, connect ideas, and apply lessons beyond the book.
Market Challenge
The book market has moved toward convenience — audiobooks, summaries, speed reading — but that convenience has come at the cost of comprehension and reflection.
Bemis addresses a gap in the market: tools that don’t just make books easier to consume, but easier to understand, engage with, and act on.
Build Challenge
- Designing a mobile-first experience that feels natural for both reading and listening contexts.
- Creating seamless transitions between listening, questioning, and note-taking.
- Avoiding the “generic chatbot” pitfall by grounding every conversation in the structure of the book.
- Ensuring low-latency responses so the interaction feels like dialogue, not a search engine.
- Building lightweight flows that work whether a user starts with a specific book or simply a curiosity about a topic.
Approach
I designed Bemis by blending familiar patterns with new interactions. Book covers, highlights, and notes anchor the experience in what readers already know. AI adds a conversational layer that encourages reflection without breaking flow.
- Structure: Each book page follows a clear flow (overview → insights → discussion → notes).
- Conversation First: Users type or speak; AI listens and guides, always tied back to the book.
- Voice Mode: Optimized for audiobook listeners who want to pause, ask, and reflect without leaving the listening context.
- Notes Integration: Conversations automatically summarize into editable notes, which can sync to Notion for extended use.
Bemis isn’t about replacing books. It’s about making them more alive — extending engagement, deepening understanding, and turning insights into action.
Results
Bemis is still in beta, so results are focused on early signals rather than long-term outcomes.
- Engagement Depth: Early users averaged 8+ back-and-forth exchanges per book session, showing demand for reflective dialogue.
- Note Retention: Over 60% of sessions resulted in at least one saved or edited note.
- Book Engagement: Users bookmarked 2–3 books on average for later exploration, extending their time with content beyond a single session.
Lessons Learned
Readers want structure. Grounding conversations in chapters and key insights worked far better than open-ended prompts.
Voice drives deeper engagement. Users who started in voice mode stayed longer and captured more notes.
Notes are the bridge. Editable scratchpads became the most valued feature, turning passive insights into personal takeaways.