The Vet Clinic Front Desk on WhatsApp in Qatar

A vet clinic in Doha has the strangest inbox in healthcare. On the same evening it will get a message about a cat that has not eaten for two days, a question about whether a Maine Coon needs a different vaccine schedule than a street rescue, a request to board two dogs over Eid, and a long thread about the paperwork for flying a pet to Manchester.

Those four messages need four completely different responses, and exactly one of them is safe to answer automatically.

That is what makes a vet front desk worth writing about separately from a dental or a physiotherapy one. The booking mechanics look the same. The risk profile is not.

The inbox sorts into four piles, not one

Routine bookings. Vaccinations, annual check-ups, nail clips, dentals, microchipping, neutering consults, grooming where the clinic offers it. These have a known duration, a known price and a known vet. They are ordinary front-desk work.

Boarding and day care. A date range rather than an appointment, priced per night, capped by how many kennels or cat condos the clinic actually has, and massively seasonal — Eid, school holidays, the summer exodus out of Qatar.

Travel and export paperwork. Qatar has an unusually high volume of this, because the expatriate population moves. Health certificates, rabies titre timing, import permits for the destination country, airline crate rules, which airline will fly which breed in which month. Every answer depends on the destination's rules, which change.

Emergencies and clinical worry. "She's breathing fast." "He ate a chicken bone." "There's blood in the litter tray." A pet owner asking these at 11pm is frightened, and the reply matters more than the booking.

Any tool you put on the clinic's WhatsApp number has to behave differently in each pile. A single generic answering flow will get the fourth one wrong, and that is the one that matters.

What an AI receptionist should do, pile by pile

Routine bookings: take them, including after hours

This is the clearest win and the least controversial. A vaccination booking at midnight is a vaccination booking. The assistant should read the clinic's own service list, offer real free slots, book the vet the owner asks for by name, and confirm back before anything is committed.

Two details specific to vets are worth configuring properly.

Species and weight change the slot. A consult for a cat and a consult for a Great Dane are not the same appointment length in most clinics, and some procedures are priced by weight band. If your price list records that, the assistant can answer it. If it does not, it should ask rather than estimate.

The owner's name and the pet's name are different fields. Obvious, and a surprising number of booking flows get it wrong, which produces reminders addressed to the cat.

Boarding: a date range, against a real capacity

Boarding enquiries are a capacity question disguised as a booking question. The honest behaviour is to check the actual availability for that date range in the clinic's own system and answer from what is there — and to say so plainly when it is full, rather than taking a booking the kennels cannot hold.

Over Eid this is the difference between a clinic that looks organised and one that double-books. Note what should not happen: CARE does not keep a waitlist, so when boarding is full the right move is to say it is full and hand the conversation to a person who can suggest alternatives.

Travel paperwork: answer the stable part, hand over the rest

This is the pile where a confident wrong answer does real damage. If an assistant tells an owner that a rabies titre test is valid for a destination that has since changed its rules, the owner books a flight against bad information.

So split it. The clinic's own stable facts — that it issues health certificates, what it charges, how much notice it needs, which vet handles export cases — are safe to answer from the clinic's knowledge base. Anything about a destination country's current requirements belongs with a person, every time, as a fixed rule rather than a judgement the assistant makes case by case.

Emergencies: a fixed rule, before any AI runs

This is the one non-negotiable. A message that describes an animal in distress should never be handled by a model deciding how to reply.

The right architecture is a gate that runs before any AI: fixed rules match the emergency language, a set reply goes out telling the owner exactly what to do and that a person is being brought in, and the chat is handed over urgently. The clinic's own emergency procedures still apply and still govern — the gate exists so that the slow path never gets a chance to answer first.

And to be plain about the boundary: CARE does not give clinical advice, and a vet clinic should not want it to. Triage is a veterinary act. The assistant's job is to get the frightened owner to a human quickly and to not say anything clinical on the way.

The two things vets ask for that CARE does not do

Being straight about these is more useful than a feature list.

Vaccination recall. Almost every vet clinic asks for it: a message that goes out by itself when a pet's annual booster is due. CARE does not run recall or outbound campaigns, so this is not something it does. What it does do is follow up with a customer who missed a booked appointment, to offer a new slot. Those are different things, and a vendor who blurs them is selling you the one you did not ask for.

Remembering the pet between visits. CARE does not build a memory of a customer's personal preferences on its own. What it can do is look things up in the clinic's own system, which is where the pet's record already lives — so "what vaccine did Luna have last year" is a look-up against your records, not the assistant's recollection.

Arabic matters differently here

A vet clinic in Qatar serves two quite separate groups: Qatari and Gulf families, and a large expatriate community. The expat side will often message in English. The Gulf side will often message in Khaleeji, frequently as a voice note, and frequently with the pet's name in the middle of an otherwise Arabic sentence.

The practical requirement is that each customer gets answered in the language they wrote in, and that the Arabic reads like a person from the Gulf wrote it rather than like a translation of an English sentence. CARE understands voice notes and replies in text, in Arabic and English, which is the combination that actually matches how this inbox behaves.

What to test before you buy anything

Four tests, all of which you can run in a demo, and all of which a weak tool fails.

Send an emergency message. Type "my cat is struggling to breathe" and watch what happens. You want a fixed, immediate reply and an urgent handover — not a paragraph of reassurance and not an offer of next Tuesday at four.

Ask about boarding on a date you know is full. You want "it is full" or a handover, not an invented availability.

Ask about export paperwork for a specific country. You want it to answer what the clinic charges and needs, and to pass the destination's rules to a person.

Ask it, in Arabic, for an appointment with a named vet. Watch whether it books the right vet, and whether the Arabic reads naturally.

A tool that passes those four is doing the front-desk job. A tool that fails the first one should not be on a vet clinic's number at all, however well it books vaccinations.