Ikigai Book Analysis
A reading companion that turns the Ikigai book into a guided, interactive experience — chapter summaries, source-cited Q&A, and your own highlight library. Web is live; mobile is shipping next.
Open live appFrom logotherapy to ikigai
Viktor Frankl wrote that the meaning of life is found in the work we do, the people we love, and the suffering we endure with dignity.Ikigai narrows this further: the reason you get out of bed in the morning.
The intersection lives where four circles overlap — what you love, what you’re good at, what the world needs, and what you can be paid for. Most lives miss one circle; the long ones find all four.
The premise
Most people read a self-help book once and forget 90% of it. Ikigai — the Japanese book about purpose and longevity — deserves better than that. The app turns the book into something you can live with: a reader that remembers where you left off, a chat that answers in the book’s own words, and highlights that follow you across devices.
The web app is live now. A React Native shell is in flight to make the same library available on the phone, with the same highlights and the same chat history.
What it does
- Chapter-aware readerRead the book chapter by chapter, with reading time, progress, and a sidebar that remembers where you stopped.
- Ask the bookA retrieval-augmented chat that answers from the book itself. Every answer cites the chapter and section it pulled from.
- Your highlight libraryHighlight a passage and it lands in a personal library — tagged as insight, principle, or practice — synced across devices.
- Daily practice promptsShort, book-grounded prompts that turn ideas like flow and small daily duty into something you can try today.
- Mobile parity (soon)React Native shell with offline reading, native share to journals, and streak tracking for daily practice.
- Source-of-truth answersNo hallucinated quotes. The model only ranks and rephrases passages that actually exist in the indexed chapters.
Ask the book
Every question is answered against the indexed chapters. The model ranks passages by relevance, then writes a short answer in the book’s voice and cites the chapter it drew from — so the reader can verify, not just trust.
- youWhat does the book say about retirement?
- ikigaiJapanese has no word for "retire" in the western sense. Garcia and Miralles’ interviews in Ogimi suggest centenarians keep small daily duties — a garden, friends to greet, a craft to repeat. The advice: never fully stop. Slow down, but don’t step away.ch.06 · centenariansch.02 · antiaging
- youGive me one practice I can try today.
- ikigaiPick one craft you can repeat without screens for 20 minutes. Cooking, walking, sketching. The book calls this entering "flow" — the unbroken focus that lets meaning settle in.ch.04 · flow
How a question becomes an answer
The book is chunked at section boundaries, embedded once, and stored in a vector index. At question time the top-k passages get retrieved, stitched into a prompt, and answered with citations attached.
The retrieval prompt
The model is told it is the book itself. It can only answer from the retrieved passages, must cite them, and must say so plainly if a question falls outside the book.
const passages = await store.search(question, { k: 4 });
const grounded = passages.map(p =>
`[${p.chapter} · ${p.section}] ${p.text}`
).join('\n\n');
const reply = await llm.complete({
system:
'You are the book "Ikigai". Answer only from the passages below. ' +
'Cite the chapter for every claim. If the answer is not in the ' +
'passages, say so plainly — do not guess.',
prompt: `Question: ${question}\n\nPassages:\n${grounded}`,
});
return { text: reply, sources: passages.map(p => p.citation) };Coming to your pocket
The mobile app is the same product, in a native shell. Same login, same library, same chat history — plus offline reading, daily practice streaks, and a share sheet that drops highlights straight into journal apps.
- Same highlights · same chat history
- Offline-first reading mode
- Streaks for daily practice
- Native share to journal apps
Stack
Web is Next.js on Vercel. The AI layer is retrieval-first — embeddings in pgvector, generation only on top of retrieved passages. Mobile reuses the same API, so it ships with zero new backend.
- 04layerMobile (coming soon)Same API, native shell. Offline reading, native share.
- React Native
- Expo
- AsyncStorage
- 03layerWeb appReader UI, chat, highlights. Fast and route-cacheable.
- Next.js
- React
- Tailwind
- 02layerAI layerRetrieval over chapter chunks. Answers cite the source.
- OpenAI
- Embeddings
- Vector store
- 01layerEdge + dataServerless functions, persisted highlights, identity.
- Vercel Edge
- Postgres
- Auth.js
What ships in v1
- 01 · read
Chapter-aware reader
Seven chapters, reading time, progress tracking, a sidebar that picks up where you left off.
- 02 · ask
Source-cited chat
Retrieval-augmented Q&A. Every answer carries the chapter and section it came from.
- 03 · keep
Personal library
Highlight, tag (insight · principle · practice), and revisit. Synced across devices.
- 04 · practice
Daily prompts
Book-grounded micro-practices — pick a craft, take a walk, find a small duty.
- 05 · share
Mobile shell (next)
React Native app with offline reading, streaks, and native share targets.
Where it stands
The web app is live at the URL above and ready for first readers. The mobile shell is in active development behind the same API contract, so day-one parity is the goal — not a stripped-down follow-on.