Shipped Product

Deejiar, live on the App Store

Roles
Product · Design · Engineering
Go-to-market
Tech
Client side: SwiftUI (iOS), Vue (web)
Backend: FastAPI, Supabase, Resend
Timeframe
May 2026, San Francisco and Tokyo
Founder @ Deejiar

Deejiar is an iOS app for planning trips, with an AI butler and a map of hand-picked places in Tokyo, Singapore and Taipei. I took it from idea to App Store: product direction, design, engineering and go-to-market.

Built for the new affluent

A new class of travelers can afford a great trip but can’t afford to waste one: a long flight, limited time off, no second try.

  • Travelers a year

    ~5M

    English-speaking, aged 20–40, household income above $100K, visiting Asia's top 10 cities.

  • Spend per trip

    $3,000–5,000

    On what they can book, before flights.

  • Bookable market

    ≈$15B a year

    5M travelers at the low end, $3,000 per trip.

My top-down estimate from tourism-board and industry reports, not yet validated with these travelers.

What the market gives them

  • Travel advisors

    Trusted and personal, but priced for the top of the market.

  • Reviews and rankings

    Open to everyone, and 54% of UK travelers worry they're fake or AI-generated.

  • AI planners

    Instant and nearly free, but a guess about a place it has never been.

Icelandair survey of 2,000 UK travelers.

What’s missing is the middle: a day shaped by someone they trust, at a consumer price. In the same survey, 58% now rely on word of mouth from family and friends.

The best journeys aren't planned. They're inherited.

Built around one move: inherit a day

Take a day someone shaped and make it yours. Everything else in the app either starts a day when there’s none to inherit, or carries it through the trip.

Text to update your journey

Tell Deja what to change. You get your own copy of the day, with every change labeled and one tap to undo.

How I build with AI

AI does the middle of the work. I own both ends: what it should make, and whether it did.

Deja's body is an equation.

The lattice

y = 1 − 2(i + ½) / n
r = √(1 − y²)
angle = i · π(3 − √5)

Every dot’s position comes from one rule. The golden angle spreads the dots evenly, so no seam or pole gives the sphere away.

Drawn by hand, every size and frame would be its own artwork; here one rule runs live at any size. AI let me work at that level: I described a behavior, read the result and changed the numbers. Where math has no opinion — the eyes, the expressions, a five-second Thinking loop choreographed in Figma — I drew it by hand. The math decides the body. I decide the personality.

Golden angle

Each dot turns 137.5° from the last.

Sphere

300 dots, evenly spread.

Depth

Size and opacity only, no shading.

Eyes

Two ovals, sized as ratios of the radius.

Dot lattice adapted from thinking-orbs (MIT), by Jakub Antalik and Alex Brinza.

AI development

Judgment sits at both ends: /plan says what to build, /inspect checks what came out, and /build in between may only stop and ask.

The nine commands are open source: coding-agent-workflow on GitHub

Choosing the model behind Deja

Deja, Deejiar’s in-app butler, turns a line like “4 days in Taipei in November, love food” into a day-by-day Journey. It’s told to build from Deejiar’s own curated places and search the web only when those run thin. So a better model has to be better at that, not just newer.

I measured it as the Deejiar share: the percentage of a Journey’s stops that are real Deejiar places. The bar is 70% in Taipei, where coverage is strongest.

The model proposes; code decides.

The model proposes · OpenAI

Ask, then draft

Asks where, how long and what you like, then drafts the days and Stops.

Edit by name

Changes a Journey through named edits (add, move, replace a Stop), never by rewriting it.

Search when thin

Uses the web only where Deejiar’s own places run out.

Code decides · Python

Places are checked

A Stop keeps a Deejiar place only if this turn’s facts carried it; the rest are matched by name, then pinned with Google.

Dates and limits

The server sets every date; at most 7 days and 8 Stops a day. A call that breaks a rule is refused, with no Journey and no credit.

Copies, never overwrites

A visitor’s change lands on their own copy; the source Journey is never written.

Five requests, one harness

Every model got the same five requests through one harness: the production planning turn, sent straight to the model at its default effort and scored by the app’s own plan-building code. gpt-5.4 scores the same through the real API route, so the harness is faithful. Tokyo and Taipei pit Deejiar places against the web, Lisbon has no Deejiar coverage, Kyoto gives no dates, and a vague “somewhere in Asia” should end in questions. A run passes when every request ends in questions or a Journey, Taipei reaches 70%, and every Journey is private.

Model harness · each model at its default effort · model seconds
gpt-5.4default (none)gpt-6-astradefault (medium)Sonnet 5.5default (high)Opus 5.5default (medium)Haiku 5.5default (medium)
Tokyo100%7.5 s55%23.8 s29%32.6 s75%38.7 s22%19.7 s
Taipeibar: 70%100%10.9 s75%39.6 s100%21.3 s100%23.0 s100%19.8 s
Lisbon0% expected0%9.5 s0%26.2 s0%24.2 s0%31.8 s0%17.9 s
Kyoto100%5.5 s83%26.1 s88%19.0 s75%34.4 s83%11.5 s
Vague64%9.2 s33%28.4 s80%23.0 s50%22.5 s78%20.5 s
Result5/543 s5/5144 s5/5120 s5/5150 s5/589 s
Model harness · each model at its default effort · model seconds
TokyoTaipeibar: 70%Lisbon0% expectedKyotoVague
gpt-5.4default (none)5/5 · 43 s100%7.5 s100%10.9 s0%9.5 s100%5.5 s64%9.2 s
gpt-6-astradefault (medium)5/5 · 144 s55%23.8 s75%39.6 s0%26.2 s83%26.1 s33%28.4 s
Sonnet 5.5default (high)5/5 · 120 s29%32.6 s100%21.3 s0%24.2 s88%19.0 s80%23.0 s
Opus 5.5default (medium)5/5 · 150 s75%38.7 s100%23.0 s0%31.8 s75%34.4 s50%22.5 s
Haiku 5.5default (medium)5/5 · 89 s22%19.7 s100%19.8 s0%17.9 s83%11.5 s78%20.5 s

Share of each Journey's stops that are Deejiar places, one run per cell, with that request's model time beneath.

Still gpt-5.4, for now

Speed × token cost

Token cost per Journey$0$0.03$0.06$0.09$0.120 s30 s60 s90 s120 s150 sModel time, five requestsgpt-5.4 · default (none)43 s · $0.042Sonnet 5.5 · default (high)120 s · $0.055Opus 5.5 · default (medium)150 s · $0.096Haiku 5.5 · default (medium)89 s · $0.003Token cost per Journey$0$0.03$0.06$0.09$0.120 s30 s60 s90 s120 s150 sModel time, five requestsgpt-5.4 · default (none)43 s · $0.042Sonnet 5.5 · default (high)120 s · $0.055Opus 5.5 · default (medium)150 s · $0.096Haiku 5.5 · default (medium)89 s · $0.003
gpt-5.4 is the fastest by 2×. Sonnet 5.5 costs about a third more; only Haiku 5.5 is cheaper, at a far lower Tokyo share.All five models passed at default effort. Token cost only; web-search fees excluded. gpt-6-astra isn't plotted: it has no list price yet.

One forced create_journey call

The same request to every model, timed end to end

010203040gpt-5.43.4 sgpt-5.55.5 sgpt-6-luna14.7 s1,437 reasoning tokensgpt-6.1-sol24.8 s516 reasoning tokensSeconds010203040gpt-5.43.4 sgpt-5.55.5 sgpt-6-luna14.7 s1,437 reasoning tokensgpt-6.1-sol24.8 s516 reasoning tokensSeconds
Default reasoning added the latency.One identical forced call per model, timed end to end, run once. Reasoning also shares the output cap: gpt-6-luna's one broken Journey was its tool-call JSON cut off mid-string.

gpt-5.4’s default effort is none: it answers without reasoning. Every other model reasons by default and took 2–3.5× its model time for the same five requests; models that searched the web more also drew fewer stops from Deejiar.

Every model passed 5/5 with no broken tool call, but none matched gpt-5.4, the only one at 100% in Taipei, Tokyo and Kyoto. Production still runs gpt-5.4; a challenger that wins here is confirmed through the real API route before it ships.

What changed is the cost of asking. The harness tests a new model in minutes, without touching production.

Sixteen releases in under five months. Speed only matters if it changes what you do next: in that window Deejiar moved from guides to travelers, from a web wrapper to native iOS, and from two Journey formats to one.

GitHub release notes for v0.03.5 — Spots are the one post type, places in your Language, first-run tutorials: 163 commits

What's unproven, and what's next

Deejiar is live on the App Store. 40 people have signed up, and a few already plan their trips in it.

Next: the first stranger who pays for Member. A friend has paid before; a stranger paying would show the product made the sale, not the friendship.

A boat crossing turquoise water, overlaid with the line: the best journeys aren't planned, they're inherited