Lorely · Independent iOS product
Building a reading product end to end
Lorely explores how readers can get useful context about a book without having future plot details spoiled. The challenge is as much about reliable knowledge boundaries as it is about generating answers.
My role Founder and sole developer — product, iOS, backend, AI and delivery
01 / Context
The challenge
Generic LLMs are good at answering questions, but readers need answers constrained to what has happened so far. A helpful reading companion must model that boundary rather than rely on a vague instruction not to spoil.
02 / Contribution
What I did
- Built the native SwiftUI app and the services that support reading and AI-assisted book context.
- Designed chapter-aware knowledge and character-specific information boundaries to avoid revealing future events.
- Own the AI processing pipeline, infrastructure, observability, deployment, subscription integration and product decisions.
- Use coding agents to accelerate development while keeping architecture and product judgement in human hands.
03 / Outcomes
What it enabled
- Created an independently developed product rather than a standalone technology demo.
- Explored a reusable approach to time-based knowledge and grounded, context-sensitive reader assistance.
- Product development is ongoing; these are architectural capabilities, not externally audited quality claims.
04 / Takeaway
What I learned
Building a product myself keeps the engineering trade-offs concrete: latency, operating costs, user experience and trust all matter more than a clever AI demo.