?
The problem
Book recommendations often feel too broad. The product needed to turn a reader request into a useful recommendation flow without making the interface feel complicated.
Back to work
BlurbBuddy
An AI book discovery tool that helps readers find recommendations by mood, topic, or similar titles instead of relying on generic category lists.
AI Discovery MVPMVP buildPublic project
Services
AI product design, MVP build, Prompted search flow, Launch polish
Stack
Next.js, TypeScript, AI integration, Tailwind CSS
↗
What changed
We made the main experience conversational: readers describe what they want, then the app gives a focused set of options they can act on.
✓
What shipped
A live AI discovery MVP with a direct prompt flow, recommendation interface, public-facing polish, and room to keep improving the matching experience.
Private details protected
The case study shows only the public product direction. User prompts, private experiments, analytics, and internal product notes are not exposed.
Solution timeline
- 1
Clarify the user promise
Reduced the idea to the core reader action: ask for a book in natural language and get a better starting point.
- 2
Shape the AI interaction
Designed the prompt flow and response structure so the output felt useful rather than like a generic chat window.
- 3
Build the MVP
Created the public app experience, connected the AI flow, and kept the interface fast enough for trial use.
- 4
Prepare for feedback
Left the product in a state where real readers can try it and the next iteration can be based on observed behaviour.
Safe public proof
This page talks about the build in plain English. It avoids sensitive data and uses anonymised framing whenever the real client system is private.
Ready to build something useful?
We can turn the messy version of the problem into a practical first build, then keep the private parts private.