75 questions answered in full, including the ones that argue against us: what you lose by migrating your number, when Hachi is not worth it, and the case in which a €29 tool is the right decision.
No question matches that text.
Write to us and we will answer it: contact form.
The basics: what it does, what it does not do, and whether it fits your business.
An AI assistant that answers your WhatsApp around the clock, checks your real calendar and books the appointment. Then it sends the reminders, and if someone asked and went quiet, it writes back.
The difference from most of the market is one verb: it does not only converse, it also executes against your calendar.
Not in the usual sense. A menu chatbot makes you press "1 to book, 2 for prices" and breaks the moment you type something it was not expecting. Hachi understands free language, in text and in voice notes, and when a conversation turns delicate it flags someone on your team instead of improvising.
No, and we do not sell it that way. It takes the repetitive part off their plate — prices, opening hours, location, parking, what to do beforehand — which is 80% of the messages. Your team spends its time on whoever is actually in the building, which is where the difference shows.
Yes. Hachi is not built for a sector, it is built for a problem: people message you when you cannot answer, and by the time you do it is too late.
It works for hair and barber shops, tattoo studios, vets, garages, academies, photography, consultancies and practices. The vocabulary changes — where it says "patient" it will say "client" — and so does your list of services. Nothing else.
That is precisely where it shows most, because there is no front desk to absorb the messages: either you answer or nobody does. The Solo plan (€149/month) exists for exactly that — one calendar, one number — and there is no minimum below which we turn you away.
No. We invoice sole traders against their own tax ID without any problem, and the invoice carries Spanish VAT and is deductible like any other supplier's.
We would rather say it than sell you something you will not use. It does not pay off if you do not work by appointment, if you receive very few messages a month (losing three messages does not cover a subscription), or if WhatsApp is not your channel because you sell in a shop or over the phone.
Aesthetic medicine — the main one — dental practices, spa and wellness, and other health and therapy practices: physiotherapy, nutrition, psychology, podiatry, speech therapy. Also solo professionals and appointment-based businesses outside healthcare. The list is where we have flying hours, not who we accept.
All figures in euros, VAT not included, and no lock-in.
Five plans: Solo €149, Essential €390, Professional €690, Complete €990 and Multi-site from €1,690 a month. No lock-in: the fee is monthly and you can leave whenever you want.
They are laid out in detail on the pricing section.
Yes, a one-off charge from €290 to €1,690 depending on the plan — under two months of fee in every case. It covers loading your services, prices and durations, connecting your WhatsApp number and your calendar, configuring your real opening hours, tuning your brand's tone and training your team.
It is not charged until you decide to continue after the free trial, and it is waived entirely if you pay a quarter up front.
At the end, and only if you say yes. In order: the demo video call costs €0, configuring your assistant €0, the 7-day trial €0. If you decide to continue, that is when setup (once) and the first monthly fee appear. If you decide not to, €0. No card is asked for to run the trial.
At a €200 average ticket and a 65% contribution margin: 2 appointments for Solo, 3 for Essential, 6 for Professional and 8 for Complete. At a €400 ticket, halve each figure.
Margin is counted rather than revenue on purpose: a €200 appointment is not €200 of profit, because it consumes product and consumables. You can run the maths on your own numbers in the ROI calculator.
One customer talking to you across 24 hours, not each message. If somebody sends thirty messages in an afternoon sorting out their questions and booking, that is one conversation.
Which is why the allowances go further than they look: a clinic receiving "20 messages a day" usually lands at 150-250 conversations a month, not 600.
The service is not cut off. No customer of yours goes unanswered for having hit a cap: that would be the opposite of what you bought.
Conversations beyond the plan are billed at €0.25 each and call minutes at €0.20, itemised line by line on the invoice. If you go over two months running we tell you and suggest moving up a plan, which is almost always cheaper than paying the excess.
It is not an artificial cap to push you upmarket. Every reply consumes AI processing charged by usage; the WhatsApp provider charges per message; and there is hosting, history storage and human maintenance work behind it.
One detail almost nobody states correctly: replying within 24 hours of the customer's message costs Meta nothing. What is paid for are the messages we initiate — reminders and campaigns. The limit sizes the service; it does not cover a Meta tariff.
Yes, up or down whenever you like; it applies from the next invoice. Not getting it right at first is normal: we start from your estimate and after two months there is real data from your business to tune it.
No. The fee is monthly and cancellation is requested before the renewal date, with no penalty. There is a discount only if you choose to pay quarterly.
Credit or debit card, SEPA direct debit or PayPal, always in euros: no currency conversion fee and no international charge appears on your statement.
You receive a deductible invoice with Spanish VAT, made out to your company or your sole-trader tax ID, issued automatically each month. If you have an intra-EU VAT number, the payment gateway applies whatever is due once you enter it.
Yes, that is the Multi-site / White Label plan, from €1,690/month. Several sites or businesses in one account without mixing, the dashboard and communications under your brand, unlimited custom agents and priority support. You decide what to charge each of your own clients: the margin is yours.
The final price depends on the number of sites and the volume, so it is worked out case by case on a call.
Two different things, and both are free. It is the most common confusion.
The demo is 20 minutes on a video call, watching it with us. The trial is 7 days using it yourself, with your real customers. First one, then the other, and the first payment only if you decide to stay.
Free and with no commitment, and no payment details are requested. Nor for starting the 7-day trial.
Twenty minutes. We show you Hachi handling a WhatsApp conversation, checking the calendar and closing a real appointment, and sending a reminder. We answer your questions and work out which plan fits you.
It really does. Beforehand we set your assistant up with your services, your prices, your number and your calendar — 3 to 5 working days — and for those 7 days it handles your actual customers. You see every conversation, every appointment booked and every reminder sent.
Seven days with your business genuinely running is enough to see it clearly. And we will be straight about the reason: we do not give you a toy demo, we build the whole assistant, and that is days of our team's work before the trial starts. A free month would mean doing that work for nothing many times over, and we would end up covering it by raising the price for everyone.
Nothing. Not the configuration, not the setup fee, not the subscription. Nothing is charged until you decide to continue.
From the contact form on this site, or by messaging on WhatsApp. Outside business hours the assistant itself handles you and books it — which is, incidentally, the best demonstration available.
The part the industry explains worst, and the one that causes the most grief when it surfaces late.
They are two different products. The app — the free one on your phone — is built for a business that replies. The API is Meta's official route for companies that also message first and work at volume.
The whole thing is laid out in WhatsApp Business vs API.
Because the app is not built for starting conversations. Once you begin sending messages to people who did not write to you that day, or many in a row, WhatsApp reads it as spam: first it throttles your sending, then it suspends the number temporarily, and if it repeats, it blocks it permanently.
And that number is usually the one you have had for years, the one on your Google listing, your website and your cards.
Yes, but it is worth deciding deliberately, because one Meta rule governs everything: a number can be on the WhatsApp Business app OR on the API, never on both at once.
There are two paths and they are set out in the next two questions.
A new SIM for the API. Your existing WhatsApp Business stays untouched on the phone, with all its chats and contacts. Zero risk, and ready in the same time frame as the rest of the setup.
The catch: it is a different number from the one on your website, your Google listing and your cards, so you have to start promoting it or set up a redirect. It is what we recommend to almost everyone, especially if your current WhatsApp has years of history in it.
Yes, and this is what almost nobody mentions until it is too late: when a number moves to the API, the chat history and contacts living in the phone app stop being accessible. The API starts clean. That number stops working in the app, and with it goes access to those conversations from the new console.
Anyone who tells you "don't worry, we'll just connect your WhatsApp Business" is not telling you this. You would find out the day you go looking for a customer conversation from eight months ago.
Yes, but it has to happen before migrating and it is separate work: extracting the contact list and conversation history, organising them and loading them into the new system.
That is not automation, it is database work: done case by case depending on how many contacts and how much history you have, which is why it is quoted separately after looking at what is there. We tell you before we start, never after.
Two questions:
Meta allows free messaging to a person for the 24 hours after their last message. After that you cannot send free text: only a template approved by Meta in advance.
Day to day: someone messages at 22:00 and Hachi replies with complete freedom for the next 24 hours. If they do not write again for three days and you want to follow up, they are outside the window, so the message goes out as an approved template; as soon as they reply, a fresh window opens and the conversation is unrestricted again.
No. The system knows whether the person is inside or outside the window and picks between free message and template on its own. We prepare and submit the templates you need — reminders, follow-ups, campaigns; Meta usually approves them within 24 to 48 hours.
This is, incidentally, why a home-made bot eventually fails: without the official API and template handling, the message simply does not go out.
The part that separates an assistant that converses from one that also executes.
Yes, always. It never offers a time without having looked at your real availability, and if there is no slot it proposes alternatives that do exist.
It is the question that separates the three market tiers: a drawn flow cannot check your calendar. It is not doing it badly, it has no way of doing it. The mechanism is explained in why flow-based bots fail.
No. That check is code that runs every time, not an instruction written into a prompt that the model might skip on an odd day. It is exactly the difference you are paying for against a €29 offer.
No. The details arrive in pieces — "Thursday"… "at 11"… "I'm Marta"… the phone number at the end — and an ordinary automation either starts over on each new message or creates a duplicate. Hachi understands it is still the same booking and completes it without duplicating.
It undoes itself. If rescheduling cancels the old appointment and creating the new one fails, the system reverts the change rather than leaving it half-done. Your customer does not end up without an appointment because of an internal error.
Yes, both, and the calendar ends up updated. Nobody has to finish the job by hand.
Yes, from the Essential plan: independent calendars per site or per practitioner, each with its own hours, services and durations. The assistant knows which one an appointment belongs to. Duration, preparation time and deposit are configurable per service.
Yes, across days, not just within one message. If you already gave your name it will not ask again, and if you said you were interested in a particular service it still has that twenty messages later. A menu chatbot makes you repeat everything, which is exactly what makes people give up.
Yes, from the Professional plan. If nobody picks up at the front desk, Hachi does, with a natural voice, checks real availability and closes the appointment during the call itself. Every call is logged with its transcript and summary in the dashboard.
What happens before and after the appointment, which is where the money leaks.
With automatic reminders before each appointment — say a week, 24 hours and 2 hours before; you decide how many and when. The customer confirms or cancels by replying to the reminder and the calendar updates itself.
The point is not only that more people turn up: it is that whoever cancels does so in time, and that slot can be sold again instead of being lost.
The effect attributed to reminders with confirmation sits in the 40% to 60% range. We use 40% — the low end — in every calculation we publish, and in the calculator that number is on a slider so you can drag it down and see what happens. It is an assumption, not a promise.
If someone asked about a service and disappeared, Hachi writes back a few days later with a message tailored to what they asked about. It is not a generic nudge: it picks the conversation up where it stopped. Available from the Complete plan.
It is by a distance the most expensive thing that happens to an appointment-based business: they ask the price, think about it, and never come back.
No, and it is not a sales promise: it is how the thing is built. The assistant only replies to people who wrote first, and follow-up acts solely on conversations that person opened themselves. If someone has never messaged you, Hachi has no way to contact them.
The first message always comes from your customer. Hachi resumes conversations, it does not start them. The one exception is the campaigns you launch to your own list.
Sends to your customer base using official Meta-approved templates — not irregular bulk messaging — segmentable by customer type, service or last visit. Available from the Complete plan.
And the important part: whoever replies goes straight into a conversation with the assistant, which answers their questions and books them. Sending the promotion is easy; the hard part is handling the fifty who reply at once.
Yes, on its own: interested in a particular service, appointment booked, awaiting reply, former customer. Your team can open it and see at a glance what needs attention, and those same labels are used to segment campaigns.
We lay it out in full even where it argues against us, because it is the only way for you to compare properly.
Because under the same description — "an AI assistant that answers your WhatsApp and books appointments" — sit three different technologies:
The full breakdown is in what an AI assistant costs.
Ours included. If the person selling cannot answer them concretely, you are looking at tier 1 or 2:
Because we would rather grow with customers who stay. Every plan sits below its tier's range: the most complete is €990/month when the tier-3 market reaches €1,840, and the highest setup fee is €1,690 when that range starts at €2,000.
What we will not do is compete with tier 1 on price.
If all you need is to answer four common questions and you book nothing, a drawn flow is plenty and it is the sensible call. We would say so either way — it is a cheaper tool that does a different job well.
Those tools chain fixed steps: if A happens, do B. They are excellent for mechanical tasks. A conversation with a customer is not one: there are questions, changes of mind, objections and missing details that arrive in pieces.
An unconstrained model fills in what it does not know, because that is what it was trained to do: it sounds convincing and it is false. In Hachi the answer is checked against your real information before it is sent, and if the fact is missing, it says so and offers to hand the conversation to a person.
This is not a theoretical worry: companies have been ordered to compensate customers for what their chatbot promised.
It is the most common thing we hear, and fairly: menu chatbots frustrate people. Which is why the trial is 7 days with your real services and prices, before deciding anything. If it does not convince you, you do not continue and you do not pay.
Against hiring someone to handle WhatsApp mornings, afternoons, nights and weekends, it is a fraction: a receptionist in Spain costs €25,000 to €35,000 a year fully loaded, and takes days off.
And if the numbers genuinely do not work, we say so. Run them yourself in the calculator: it is built to tell you it does not pay for itself yet, and if that is the answer, it says so.
You handle personal data and, in healthcare, special-category data. This is not a checkbox.
You are the controller: they are your customers and your data, and you decide what they are used for. Hachi is the processor: we handle that data on your behalf, only to provide the service, following your instructions.
Yes, the one required by Article 28 of the GDPR, before we start. It details what data is processed, with which providers, where it is hosted and how long it is kept. If your legal advisers need to review it or add conditions, send it over and we look at it before signing. That is routine.
No. The assistant only replies to people who wrote first. The sole exception is campaigns you launch, and there the rules are: you supply the list (Hachi does not buy, scrape or infer numbers), the consent is yours as controller of that data, and Meta also polices it by requiring an approved template. If Meta rejects it, it does not go out.
Our recommendation: only send campaigns to people who gave you their number so you could write to them, and keep a record of when and how.
It is special category data (Art. 9 GDPR) and deserves more care, not less. Which is why Hachi works with the conversation — what the customer writes to request an appointment — and does not need your clinical records: there is nothing to upload. The less data that goes in, the better, and what is stored and for how long can be restricted.
Access, rectification, erasure, objection and portability: you pass it to us and we handle it within the legal deadline, including complete deletion of their data and conversations.
Server location, the specific list of providers involved, exact retention periods and international transfers are all detailed in the data processing agreement, handed over and signed before we start. If you need to see it before deciding, ask for it at the demo.
The opposite. From the dashboard you edit how it replies, what prices it quotes, what services it offers and what it must never say. Your team sees every conversation in real time and can step in at any moment.
It speaks naturally and in your brand's tone. And the moment a conversation turns delicate — an emergency, a serious complaint, someone asking for a human — it alerts your team and hands over with the full context. Nobody gets trapped talking to a machine.
What you need to have, how long it takes and what happens afterwards.
Three to five working days from when you send your list of services and prices. The 7-day trial starts as soon as it is built.
Nothing to install — the API runs in a browser — no need to change management software, no need for anyone to watch a phone, and no need for anyone on your team to be technical.
Hachi can work with its own calendar or sync with the one you already use; we look at that in the demo, depending on your software.
No, it is included in the fee. Price changes as often as needed, adding or removing services and durations, changes to hours, calendars, practitioners and holidays, adjustments to how the agents behave, changes to reminders, and every platform improvement. There is no surprise invoice each time you change something.
Yes: prices, wording, hours and each agent's behaviour are all in the dashboard, and there is a simulator for testing any change in a practice conversation before a real customer sees it. If you would rather we did it, message us and routine changes are done within 24-48 working hours.
Only two things: custom integrations with management software that is not already supported, and bespoke development that is not part of the product. In both cases you get a fixed quote before anything is built. Plus rescuing contacts and history if you choose to migrate your number.
We monitor the service automatically and find out before you do. A serious incident — the assistant not responding, the WhatsApp connection dropping — is handled immediately during working hours. Multi-site plans have priority support.
The dashboard shows every appointment Hachi booked, every reminder sent and every conversation recovered, plus the service status. Nothing has to be taken on faith: the numbers are there.
Sales hours are Monday to Friday, 09:00 to 19:00 (mainland Spain time) and Saturdays 10:00 to 14:00. Outside those hours the assistant itself handles you and books the demo.
Write to us and we will answer it for your specific case. And if you want numbers rather than answers, the calculator runs the maths on yours.
Book a demo Calculate my return