← Back to the deck vibe · Financials · Working model · Aug 2026 · Confidential

The numbers behind the story.

A live model, not a spreadsheet screenshot. Every input below is editable; the tables and the chart follow. Two presets: what we think will happen, and what happens if it works.

The round.

$500k

Pre-seed, post-money SAFE

$10M

Valuation cap

Ownership at the cap

—

Months of runway on the base case

Where this sits: the median post-money SAFE cap for rounds between $250k and $500k was $10M in mid-2025 (Carta, State of Pre-Seed Q2 2025). We are asking the market price, not a premium. The honest tension is runway, not price — $500k at our burn is about a year, and the App Store launch, the first paying cohort and a Series A conversation all have to fit inside it. A $750k close at the same cap (7.5%) buys eighteen months and a safer Series A timeline; $500k is the floor, not the target.

Source: Carta, State of Pre-Seed Q2 2025.

What we assume.

Three inputs do almost all the work: how many visitors we reach, how many of them buy a week, and how many locals subscribe. Change any number and everything below recomputes.

Market inputs from the Japan Tourism Agency 2025 final survey (20.9M foreign visitors to Tokyo, ¥3.29T spent there). The install rates assume a creator-led launch: Kole’s organic media network delivers over a million views a post, which is why growth spend stays small and marketing cost per paying customer stays low. Costs from production: one Ask costs ¥4 warm; writing up a place costs about ¥42. Apple's cut is 15% under the Small Business Program until $1M a year, then 30%. Exchange rate ¥149.6 = $1 (2025 average).

Three years.

Year 1 starts at App Store launch. Cities are counted as "Tokyo-equivalents" of reachable visitors; Osaka and Kyoto together are about one Tokyo.

Month by month.

Revenue after Apple against total spend, and the cash line with this round in the bank. Where the cash line crosses zero is when the next round has to be closed.

Net revenue / monthSpend / monthCash

One customer.

—

Contribution from one Trip Pass, after Apple and model cost

—

Lifetime contribution from one vibe+ subscriber

—

Gross margin on a paying user

—

Marketing spend per paying customer in year 1

Inference is the only input in this business that gets cheaper every year: the model class behind Ask cost $8/$24 per million tokens in 2023 and $2/$10 since June 2026 (Anthropic list prices). A visitor spends about ¥229,000 in Japan; $10.99 to know where to go is a rounding error on the trip (JTA 2025).

What the money does.

The product is built. The money buys reach: creators and inbound partners, the first paying cohort in Tokyo, and the next two cities.

60% growth — creators, inbound partners, the first paid cohort30% engineering — one hire, infra, App Store work10% venues — partnerships and the private-access tier

By month 3

App Store live with Trip Pass and vibe+ on sale. Osaka written.

By month 6

First 1,000 paid trips. Import (reels and Maps links) shipped. Kyoto live.

By month 9

Three cities live. Retention curve flat at six months for subscribers.

By month 12

$30–50k a month in revenue, a Series A conversation on real cohorts, New York written.

What could go wrong.

Google

Ask Maps reached Japan in August 2026, in English and Japanese. It answers from the same reviews people trust half as much as they did five years ago. If Google builds a real taste layer and a friends layer, our window narrows. Our answer is the corpus and the community, not the chat box.

Conversion is unproven

We are pre-revenue. Every paid number here is a benchmark, not a cohort. Lifestyle apps convert 2–9% of active users to paid (Strava, Duolingo). We assume 3% of locals and 8% of visitor installs. The first three months after launch tell us which end we are at.

Apple

App review, the 30% cut above $1M, and the rules on what a subscription may gate. Payments are wired through StoreKit before submission.

Freshness

A written corpus goes stale. Closures, moves and menu changes are caught by a weekly sweep and by people who go; keeping 1,700 places true is a standing cost, not a one-off.

Team size

Three people. The plan hires one engineer in the first six months; the raise size decides whether that is comfortable or tight.

Yen

Costs are in yen, prices are in dollars. A stronger yen raises our cost base in dollar terms; a weaker yen makes Japan cheaper and busier for visitors.

If it works.

What a $500k cheque at a $10M cap looks like in three cases. Multiples are on ownership at the cap, before later dilution.

It doesn’t

Launch lands, conversion sits under 2%, we don’t reach a Series A. Capital is lost; the corpus and the brand have some acqui-hire value to a booking rail or a map.

Base case

—

Upside

—

Comparables

Letterboxd — the taste-community model this is built on — sold 60% at a $50–60M valuation in 2023 with 10M members, and was in sale talks at around $250M in 2026 with 30M. Corner raised $3.75M then filed for a further $8.4M for a US-first social map with about 55k users. Amex paid $400M for Tock. Taste plus a booking rail is where the exits have been.

Sources: Variety, 2026 · Tiny, 2023 · AlleyWatch, July 2026 · Restaurant Business, 2024.