Business plan
The pet health record that fills itself in.

TailStory is the pet health record that fills itself in. A photo of the animal becomes a structured profile — species, breed, coat, body condition, size, visible health observations — in one AI call. From there the product follows a single rule: never ask the owner to supply data; infer it, then let them confirm in one tap.

The product is live in beta today, in five languages, on iOS and Android; the store releases ship with the Netherlands and Austria launch. Every screen shown to an investor is the real application: photo-to-profile, one-tap and voice logging, a medical timeline, the EU pet passport, expiring share links for a vet and a sitter, an AI vet assistant grounded in the animal's own history, and a shop that compares prices across five retailers. Built end to end by the two founders, no agencies.

What the record then does

• It answers. TailStory Doctor is an AI vet assistant grounded in that pet's record and full medical history, with retrieval over a corpus of veterinary and animal-behaviour books — the book cited under the answer.
• It observes. Health intelligence (Section 4): a nine-area photo check read by unit-tested clinical tables rather than by a model, lab reports scanned into charted values, and a trend engine. It never diagnoses.
• It travels. The EU pet passport is stored, and the app mints two expiring share links: an Emergency Pass for a vet and a Sitter Digest for whoever is feeding the cat. Both print to one A6 card.
• It buys. Per-pet product picks with a stated reason, price-compared across Amazon, Zooplus, Bitiba, Fressnapf and Maxi Zoo, the purchase completed on the merchant's own site. TailStory earns a commission, never handles the payment.

Three revenue lines, one asset

The problem, stated for an owner rather than an investor

A pet's life is recorded in five places at once: a paper EU passport in a drawer, a photo of a vaccination card, a vet's own system the owner cannot read, a message to the sitter, and memory. Nothing connects. So at the vet the owner reconstructs eight months from memory; at the border the family finds out the rabies shot expired last month; when they travel the sitter gets a hurried voice note and guesses; in the shop they buy on a photo of a bag they liked; and at 2am, when the dog is limping, they read a forum written by strangers about a different dog.

The solution, one line per problem

This section is not a roadmap. Everything in it is built and running.

What is shipped, and what each part is worth commercially

The stack, in one paragraph

An Expo / React Native app on iOS and Android talks only to a modular NestJS API which holds the privileged database key — that key never reaches a device, so every business rule lives in one place. AI is a single module in front of one provider, so the provider is swappable without touching feature code. Vector search runs in the same Postgres. Scheduled jobs, not request-time calls, load store catalogs, affiliate feeds and the vet directory into our own database.

Four positions enforced in code, not in a policy document

• Ranking is not for sale. What TailStory earns from a shop never affects the order products appear in — enforced server-side, disclosed in the app, and required of marketplaces by EU consumer law.
• The app never takes a payment for goods. It sends the user to the merchant. A compliance necessity and a trust asset at once.
• Never a medical claim. Every AI surface carries "not a substitute for a veterinarian"; red-flag symptoms and "see a vet now" come first; the health score is clamped so it can never read as an alarm. Where a model's own sentence is shown at all it is swept for medical claims first and dropped if it makes one — and urgency itself is never a model's to decide (Section 4).
• Sensitive data does not leak. Passport scans live in private storage reachable only through an expiring signed link after an ownership check. They never reach a share link and never reach a model prompt.

Three claims deserve scepticism in any consumer AI pitch: AI-powered, grounded in veterinary literature, and reads health from a photo. This section states exactly what each one means at TailStory, which parts run today, and which parts this round builds.

Retrieval, end to end — every stage is ours

TailStory Doctor is a retrieval-augmented system, not a prompt in front of a general model. The corpus, the index and the retrieval all live on our own infrastructure; the model provider is the only replaceable part.

Voice, in one call

Audio goes to the model as a media part: a single multimodal call transcribes a spoken sentence and structures it into proposed log entries, with no separate speech-to-text vendor, no second bill and no second failure mode. The same primitive gives the Doctor a microphone — and there the transcript lands in the composer editable and un-sent, because a mis-heard question produces a confident answer to something the owner never asked.

The record already knows what was done to the animal. Health intelligence is the layer that knows something about the animal itself — from a photograph of a symptom, from a lab report, and from the shape of the last ninety days of logs.

Two rules, and they are not negotiable

• The model observes; a table decides. A model may report that a stool is black and tarry. Whether black-and-tarry means "call a vet today" is a row in a unit-tested table in our code, not a sentence a model generated. There is no code path anywhere in the system in which a model chooses an urgency.
• The backend emits codes; the app writes the prose. Findings travel as machine codes and numbers. The five language files turn them into sentences. A health warning is never whatever a model felt like phrasing that call, in any of the five languages.

What the model is asked — and what it is never asked

The vision schema contains closed enumerations for colour and consistency, booleans for visible blood, mucus, parasites, discharge and swelling, an image-quality field, and a is this even the body part we asked for flag. The words urgency, diagnosis, cause, condition and recommendation do not appear in the schema at all — not as a property, not in a description. A photo of the floor returns a normal band and a "wrong area" code, not an error: it is a user event, not a failure.

Four bands, four actions

The trend engine

Fifteen rules over the assembled history of one animal, evaluated on a schedule. Each is a pure function of a context object, so every threshold is unit-tested and no rule can quietly disagree with another about what "this week's mean water intake" is. Six of the fifteen:

Why this is not a medical device, by construction

• No endpoint returns a cause, a condition name a model invented, or a dose — and the model's own summary is swept for medical claims before it is shown, or dropped.
• Extracted lab values are flagged only against the reference range printed on the page. We do not own a reference-range table and must not behave as though we do.
• A proposed chronic condition may only carry a name printed on the document. A condition concluded from the numbers is precisely the thing this feature must never do.
• Nothing is written to the record until the owner approves it, and every non-normal band carries the not-veterinary-advice notice.

The honest description of the whole layer is one sentence: it notices, and it tells you to ask someone who knows.

What it is worth commercially

Status — what is shipped and what this round builds

Every figure here is a planning assumption pending verification against European pet industry data — sized to be directionally defensible, not quoted as fact (Section 14).

European pet spend is recession-resistant and rising — owners cut their own spend before the animal's — and the multi-pet household is not a niche: about a third of pet-owning households have more than one animal, and they spend disproportionately.

Why the Netherlands and Austria first

Expansion sequence

The sitter marketplace expands city by city, not country by country: a national launch with no density is a launch with no bookings.

The moat, in order of durability

• Accumulated per-pet history. Not copyable, and it does not travel. Every week of logs makes leaving more expensive.
• Effort engineered out. Photo prefill, voice logging, the usual-day tap, the forgiving streak. This is a design system as much as a feature set, and it is the reason the history accumulates at all.
• The clinical tables and the corpus. Sixty-odd signal codes with their bands and escalators, fifteen trend rules with their thresholds, and an ingested, embedded veterinary library sitting in our own index. This is months of judgement written down and unit-tested — not a prompt, and not something a competitor gets by switching model provider.
• Multi-pet lock-in. Switching cost multiplies per animal — and that is exactly where the subscription sits.
• The document surface. Every Emergency Pass and Sitter Digest puts TailStory in a vet's or a neighbour's hands at zero cost.
• Regulatory data as code. EU travel rules, heat thresholds and carriage conditions encoded as data rather than prose. A copy is a research project, not a sprint.
• The provider abstraction. Adding a store is a feed, a column map and a database row. A competitor without it waits in an approval queue.

Any one of these is copyable in a quarter. The combination is not.

Line 1 monetises ownership. Line 2 monetises buying. Line 3 monetises leaving the house. They peak at different moments in a year, which is the point: affiliate spikes at seasonal and life-stage changes, sitter bookings spike in July–August and December, and the subscription is flat and compounding underneath both.

Line 1 — TailStory Plus

One pet is free forever. A second pet requires Plus. That is the correct place for the wall for three reasons specific to this product. The single-pet experience is already complete, so the free tier is a real product rather than a demo. Adding a second pet is the highest-intent moment in the app — someone typing a second animal's name has already decided TailStory is where their animals live. And multi-pet households are where retention, basket size and vet frequency all concentrate at once.

Store commission is 15% under the small-business tiers of both app stores, so net revenue is €5.09 per monthly subscriber and €38.24 per annual one. Billing is in-app purchase, which the stores require; the backend needs one check — is this user premium?

Line 2 — affiliate commission

Fully built. A tap on "Buy from Zooplus" opens our own redirect, writes a click row with our sub-id, and forwards to the merchant with the tag attached; a nightly import matches conversions back to the click. Four properties matter commercially: the app never builds an affiliate URL, so a changed tag or a suspended programme is a config edit rather than an app release; a missing tag still returns a working link, so a purchase is never broken to protect a commission; conversions are not counted optimistically, because networks confirm or reject weeks later; and reporting by source judges the AI directly — if recommendation clicks do not convert better than search, the model is decorating the shop and the AI spend should be cut. Blended commission assumption: 5% of basket. A Verified Shop subscription tier — boosted placement and its own analytics dashboard for retailers, still fulfilled through the same redirect — sits on top of this line; see the pricing note in Section 10, which applies to it as well.

Line 3 — the Partner Network

Specified in full in Section 10. The one-line version: sitters, pet hotels, shops and vets subscribe — from €9.99 a month, tiers priced by the Amsterdam pilot — for a verified profile and sponsored placement, managed from a separate TailStory Partner app — revenue that starts before a single booking button exists. TailStory already produces the document a sitter needs, and every Sitter Digest shared today is a handoff happening off-platform; a booking commission layers on top once reservations go live. Take rate on a booking: 15% from the sitter, 5% service fee from the owner, settled through Stripe Connect because a real-world service consumed outside the app is outside the app stores' billing rules.

Cost to serve one user

Revenue per acting user, per month

LTV and CAC

The organic weighting is structural, not optimistic: every Emergency Pass handed to a vet and every Sitter Digest sent to a neighbour distributes the product with no media budget.

Do not launch as "a pet app". Launch as the thing that answers one dated, frightening question. The wedge is the EU pet passport and cross-border travel: seasonal, searchable, high-anxiety, and with a deadline attached — which is precisely what makes it convert. "Can I take my dog to Italy in July?" has a date, a checklist and a real consequence for getting it wrong. From there the app opens sideways into the daily habit, and the habit produces everything else.

Channels, in priority order

Launch sequence

The third line was originally scoped as sitter-booking commission alone. It is now broader: sitters, pet hotels, shops and vets apply, get vetted, and pay a subscription for a verified profile and sponsored placement inside the consumer app — earning revenue before a single booking button exists. The sitter-booking commission on the following pages layers on top once reservations go live; it does not replace this.

What the subscription includes

Every sitter marketplace begins with a cold-start problem on both sides. TailStory starts with the handoff already in the product. The Sitter Digest exists, carries the learned usual day, the daily goals, the dose times and the house rules, expires in 14 days and is revocable. Owners are already sending it, and each one is a booking that happened without a commission.

Model

Risks specific to this line

Base case, bottom-up from the drivers in Section 8. Annual, in euro, rounded — a plan to be falsified by instrumentation, not a forecast to be defended.

Drivers

Revenue

Sensitivity — the two numbers that move everything

Cost base

Cumulative burn to end of year three is ~€1.28M. Tranche one covers months 0–11, the €250k milestone tranche to month 18; the Series-A-or-profitability decision lands in month 20.

The ask — €350,000 now, €250,000 on milestone

A €600,000 pre-seed in two tranches. €350,000 now funds months 0–11: instrumentation, TailStory Plus, the affiliate line, Partner Network subscriptions, and the Amsterdam sitter pilot. €250,000 releases on three measured numbers — multi-pet share ≥ 25%, Plus conversion ≥ 5%, 50 vetted sitters in Amsterdam — and funds Vienna and the German listing. Euro figures below are tranche one.

Team — 60% · €210,000 of tranche 1

Two engineers and one growth hire. They ship the instrumentation, the subscription and the marketplace. No agencies, and no headcount beyond these three in the eighteen months.

Marketing — 18% · €63,000 of tranche 1

The EU pet-travel content engine, vet clinic partnerships, shelter and breeder programmes, and paid social only after organic CAC has been measured.

Partner marketplace — 15% · €52,000 of tranche 1

Payments infrastructure, liability insurance, ID and business verification and vetting across sitters, hotels, shops and vets, the incident line, and supply acquisition in the first two cities.

Legal and compliance — 7% · €25,000 of tranche 1

GDPR, DAC7 seller reporting, DSA hooks, consumer withdrawal terms in NL and AT, and the contractor-classification opinions required before the sitter pilot.

Already enforced in the product

What the new revenue lines add

Top company risks

The four metrics that decide the business

Assumption register — the order to measure them in

The round funds three more people and no more: two engineers and one growth hire. Everything in the eighteen-month plan is sized to that team, which is why the plan is sequenced — instrument, charge, then open the marketplace one city at a time — rather than run in parallel.

The ask

€350,000 now, and €250,000 on milestone — a €600,000 pre-seed over 18 months: instrument the funnel, ship the subscription, launch sitter reservations in Amsterdam and then Vienna, and reach a proven € per pet per month across all three lines before raising again.