Every question, in one place

Bosun is a WhatsApp assistant for a business's own team. Staff message it the way they already message each other - "today I'm ordering the rotors and calling Mrs Diaz" - and it turns that into a tracked job list, chases what is late, and answers questions from the business's own written procedures. It is sold white-label: you run it in your brand, on your own server, for a one-time price.

No. Bosun faces inwards, at your own staff. It is not a marketing tool, a broadcast tool or a customer support inbox. Nobody outside your roster can talk to it - an unknown number is refused before anything is written down.

No. They use WhatsApp, which they already have and already open. That is the point: team software usually fails because nobody opens it, and this removes that step entirely.

No. They write the way they would to a colleague. "The Ford is stuck, we're waiting on the parts department" is understood as a blocked job with the reason recorded. The owner has a few shortcuts if they want them, but nobody has to memorise anything.

A one-time price of $1,999 - the same in every country - which includes setup in your own accounts, your branding, your WhatsApp connected, and a year of support and updates. There is no monthly fee to us. You pay your hosting provider (about $12-25 a month), your AI provider (about $5-15 a month) and WhatsApp's own message fees directly, in your own name.

Because it is your key and your account, so you see exactly what it costs and can change provider whenever you like. Measured on the running product it is about $0.002 a message: a ten-person team sending 2,000 messages a month is roughly $4. Bundling it would only mean marking it up.

Nothing stops. The software keeps running because it is on your server, in your account. Renewing support is optional and yearly - it buys fixes, security and platform updates, and help. It is never monthly.

Not on the standard licence, which covers one business. If reselling is what you want to do, that is a partner arrangement - say so on the contact form and we will talk about it properly rather than pretending the standard price covers it.

About a week from the day you send us your brand, your team and your procedures. The one part nobody controls is WhatsApp Business verification, which Meta takes one to two weeks over. We start that first, and everything else happens while it runs.

Yes, and it should be a number your team does not already use for personal chat. We wire it to the assistant and submit the two message templates Meta requires for reminders sent outside the 24-hour window.

Your business name and colour, your team's names and WhatsApp numbers, your departments, and whatever you already have written down as procedure. If nothing is written down, that is worth knowing too - the assistant will show you every question it could not answer.

Either OpenAI or Anthropic, on your own key, switchable in the admin. Both are tested before each release against the same checks. The cheaper tier of either is enough for this work.

No. You set your working days and hours, and no reminder is ever sent outside them.

No, and that is deliberate. It reminds a person a set number of times - you choose how many - then stops and tells the owner instead. A tool that nags forever gets muted, and a muted tool is worse than none.

No. Only the owner or a manager can put a job on someone else's list, and that is enforced in the code rather than asked for in a prompt. A daily plan only ever creates work for the person who sent it.

It says so, and records the question for the owner. It will not invent a spending limit or a policy. The Procedures screen shows every question your written documents failed to answer, which is usually a more useful list than it sounds.

None to us. You pay once, and the software runs on your own server for as long as you want it. What you do pay monthly goes to other companies, in your own name: hosting, your AI provider key, and WhatsApp's own fees for messages sent outside the 24-hour window. Together that is usually $20 to $40 a month.

No. Add your whole team - six people or sixty - for the same price. The running cost rises slightly with how many messages they send, because that is the AI cost, but there is no per-seat fee.

Nothing stops working. Support is optional and yearly, and it buys fixes, security and platform updates, and help. The software itself is on your server and keeps running whether you renew or not.

Because it already exists and it has been tested. Building a WhatsApp assistant that reads free text, tracks work, chases people and answers from documents is two to three months of competent development before you find the failures - and those failures are the expensive part. Three of the defects in this product were only found by running the test suites fifteen times in a row; none failed on the first pass.

Yes. Every message costs about $0.002, and the dashboard shows the spend per day and per purpose from day one. In the demo we can show you the actual figures on live traffic rather than an estimate.

No. We do the setup, and everything you are likely to change afterwards - people, departments, working hours, reminder timings, procedures - is a form in the admin. Only new code needs us, and that is quoted as a fixed price before it starts.

Then the assistant will refuse to answer those questions and record them for you, which turns "we should write that down one day" into a specific, ranked list of what your staff actually ask. Many businesses start with three or four pages and grow from there.

Yes, and you should. The dashboard has a simulator that lets you send messages as any member of your team and see exactly what the assistant does - no real messages, no cost. Most of the setup is tested that way before your number is live.

Meta usually takes one to two weeks, and it is outside anybody's control. We start it on day one so it runs alongside everything else. If it drags, everything except the live number is already finished and waiting.

It is a demonstration, and you will be talking to the person who built it. If Bosun is wrong for you, we would rather establish that in twenty minutes than after you have paid.

Yes. Send a document or two beforehand and we will load them, so the answers you see come from your rules rather than a sample dealership's.

No. It helps to bring one real example of something that got dropped recently - a job nobody followed up, a question that went unanswered - so we can show you what would have happened instead.

No. That is the whole design. They write the way they already write - "today I'm ordering the rotors and calling Mrs Diaz" - and the assistant does the structuring. There are no commands to learn and no forms to fill.

Bosun works in one-to-one chats with each person, not in your group. Group chats are where things get lost: a message scrolls past and everyone assumes somebody else picked it up. Each person having their own thread is what makes a job traceable to a name.

Both are kept. A voice note is transcribed and handled exactly like a typed message, and it repeats back what it heard - see <a href="/features/voice-notes/">voice notes</a>. A photo is stored and attached to the job the same message created, so the picture of the damage sits on the job about the damage. What it does not do is interpret what is in the photo; a person still reads it.

Anything the assistant is unsure about is flagged for the owner to review rather than quietly filed, so low confidence never becomes a confident-looking entry. Nothing is ever deleted either, so the full history of every job stays on the dashboard.

That works too - bullets, dashes, numbers or one long sentence all read the same. The formatting does not matter; the deliverables do.

No. A daily plan only creates work for the person who sent it, and that is enforced in the code. Only an owner or a manager can put a job on someone else's list, and they do it explicitly.

Restating existing work is treated as an update rather than a new job, so you do not collect duplicates every morning from somebody listing what they are still working on.

You will see it straight away, because the reply opens with what it heard. Correct it in a follow-up message the way you would correct anything else. That is why the transcript is repeated back rather than quietly acted on.

Yes - set the language in the admin and it is used for every note. The assistant already replies in the language the person wrote in; this is the same idea for speech.

Because Anthropic has no speech-to-text API. Rather than pretend otherwise, the setting says so and lets you enter an OpenAI key used only for this. If you already run the assistant on OpenAI, it reuses that key and you enter nothing.

From WhatsApp to your server, then to your own OpenAI account to be transcribed. We are not in that path - see <a href="/trust-and-security/">trust & security</a>. The audio is not kept; the transcript is, as the message.

Not yet. A photo is stored on the conversation so it is not lost, but it is not understood. Tell us on the <a href="/contact/">contact form</a> if that is what your business needs next.

It is kept, and the reply says so and asks what it is for. Send the words afterwards and the next job you create carries it. Nothing is thrown away just because it arrived without a caption.

A phone photo is one to three megabytes. A busy team sending twenty a week uses a few gigabytes a year, which is nothing on the hosting the rest of this runs on. The dashboard shows the running total.

Yes. The backup script takes the database, its encryption key and the attachments together, because a job list pointing at photos that were not backed up is worse than either alone.

Yes, from the dashboard. The file itself is only removed once no other record points at it, so deleting a duplicate never takes another job's evidence.

It behaves like any other job. Give it a deadline and it is chased - privately, inside your working hours, up to the number of attempts you allow, then escalated to you. Leave it without one and it sits on the list until somebody closes it, and shows on the owner's briefing as outstanding.

A daily rule creates a job each day it falls on, so a week of closure leaves a week of open jobs. Pause the rule while you are closed - pausing keeps everything it has already produced and stops new ones, and switching it back on resumes from that day.

A job belongs to one person, as everywhere else in the product - that is what makes it traceable. For a check anybody can do, leave it unassigned and it appears on the owner's list to hand out, or create one rule per person.

Nothing measurable. Creating the job is a database write, not a call to your AI provider, so a dozen recurring checks a day add nothing to the running cost on the <a href="/pricing/">pricing page</a>.

No, but if they do not, the assistant asks them at the end of the day - once, at a time you set, and never outside your working hours. Who has not reported also appears on the owner's briefing.

Anything it cannot confidently map is quoted back verbatim and left for the owner rather than acted on. A wrong close is worse than an unhandled sentence.

Yes - saying they have started it again moves it back, and every change is recorded in that job's history with who did it and when.

That was the main risk, and the limits above exist because of it. It messages privately, only about their own work, only inside working hours, and only a set number of times. After that it leaves them alone.

Yes - the intervals, the number of attempts and the working hours are all settings in the admin, changed by you at any time without us.

Mark them inactive in the admin and they stop receiving anything immediately. Their jobs stay on the list, visible to the owner for reassigning.

No. Reports and team-wide questions are owner and manager only, and that is enforced in the code rather than asked for in a prompt. A staff member who asks is told plainly, and the refusal contains no data.

Yes. Set a time in the admin and it arrives on WhatsApp, and by email too if you switch that on.

Identity is the phone number checked against your roster, not anything the message says. A message claiming owner privileges is refused - that specific attempt is one of the tests run before every release.

Markdown and plain text work directly, and we convert your existing Word files or PDFs during setup. Headings matter more than formatting - a document with clear sections answers better than one long page.

Every grounded answer cites the document it came from, so any answer can be checked in seconds. Before each release, every money figure the assistant produces in testing is verified to appear verbatim in the source documents.

It answers from what it finds and cites the source, which surfaces the contradiction to a human rather than silently picking one. That is usually a document problem worth knowing about.

Yes. Add or edit a document in the dashboard and it is re-indexed immediately. No developer, no re-deployment.

No. Half sentences, shorthand, no punctuation and decisions buried in prose are all normal, and all read fine. The tests for this feature deliberately use badly typed notes, because that is what real notes look like.

Not today - it reads typed notes. Transcription is the missing step, and it is on the list; tell us if it matters to you.

Yes. Send them starting with the word "meeting" and the reply tells you what was recorded, what became a job and what is still open.

The figures work from week one, but the comparison needs a second week to compare against. The recommendations get sharper after three or four weeks, when there is enough history for a pattern rather than a bad Tuesday.

Yes. Each week is stored when it closes, and the dashboard lets you open previous weeks and see the trend of created against closed.

No. It is owner and manager only, like the other reports.

Then the assistant sees one person, because the number is the identity. For a shared workshop phone, add it as one roster entry with a name everyone recognises - or better, use individual numbers, since the value is in knowing who is carrying what.

Yes. Several people can hold the owner role, and managers can do everything except change the system settings.

Nothing is deleted. Deactivate them and their open work stays on the dashboard for you to reassign, with its full history intact.

No. An ordinary Google account works. Setup uses a service account we create during installation, and the spreadsheet stays in your own Drive.

You can type anything you like, but only the Status column is read back. Everything else is overwritten on the next sync, because Bosun is the source of truth for the rest.

No. It is off until you switch it on, and it can be switched off again without affecting anything else.

Any SMTP account - your own mail host, Google Workspace, or a transactional provider. You enter the credentials in the admin and there is a test button that reports failure honestly rather than claiming success.

Yes, though it contains team-wide figures, so it is worth thinking about who sees it.

It still calls your AI provider, so it costs the same fraction of a cent as a real message - around $0.002. It sends nothing through WhatsApp, so there is no message fee.

Yes - jobs it creates are real jobs, which is what makes it a useful test. Delete them from the dashboard afterwards if you are only experimenting.

Yes, if you would rather not run a server. It is a separate arrangement with its own monthly cost - tell us on the <a href="/contact/">contact form</a> and we will quote it plainly. Most buyers self-host, because that is the point.

Nothing changes commercially - there is no per-seat price. You add people in the admin. The only figure that moves is your AI provider bill, because more messages mean more calls, at roughly a fifth of a cent each.

Yes, unobfuscated, with the automated test suites. You can read it, change it and keep running it whatever happens to us.

No, and this matters. Tools built on whatsapp-web.js or Baileys break WhatsApp's terms and get numbers banned. Bosun uses Meta's official Cloud API, which is the only route that is safe for a business number.

Meta lets a business send free-form messages only within 24 hours of the customer - here, your staff member - messaging it. Outside that, messages must use a template Meta approved in advance. Bosun tracks each person's window and picks the right kind automatically, so you never see a failed send.

Yes, and most do - that is why adoption is not a problem. If you would rather not have work on personal phones, that is a policy question for you rather than a technical one.

It answers them - they chose to. The working-hours limit governs what the assistant initiates, not when people may write to it.

Almost. Managers can assign work and pull every report; only owners can change the system settings and credentials.

Yes - create, reassign, change a deadline, change a status, add a note. Every edit is recorded with who made it, like any other change.

No, and deliberately - it is a web dashboard that works properly on a phone browser. An app would be a second thing to install and update for no benefit.

Three reasons: you pay cost rather than a marked-up subscription, your usage is never throttled by another buyer's, and nothing about your business passes through our account. It is the main reason the running cost is a few dollars a month.

Yes. OpenAI and Anthropic are both supported and the choice is a setting, so you can move if pricing or quality changes. Nothing in your data depends on which one you picked.

Then the stored credentials cannot be decrypted and you re-enter them - annoying, not fatal, since your jobs and people are not encrypted with it. The backup script takes the key and the database together so this does not come up.

No. Everything on this page is a form in the admin. You need a developer only if you want to change the software itself, which you are entitled to do.

No. It is how the product works - the branding is settings, not a service we perform. The price on the <a href="/pricing/">pricing page</a> includes it.

Not in the running product. It appears in the source code's own documentation and comments - which your staff will not read, and which you can change, since the code is yours.

Yes, any time, in the admin. It applies immediately and nobody needs to re-install anything.

Probably, if the description above fits: a team of five to fifty, work agreed in conversation, staff who are not at desks. The six pages exist because those are the trades we have written procedures for, not because the software knows the difference. Tell us what you do on the <a href="/contact/">contact form</a> and we will say honestly whether it suits you.

The departments, the starter procedures and the wording. That is it - the mechanism is identical, which is why a dealership and a clinic run the same release.

Not cleanly, no. One installation is one business - that is what keeps it simple and cheap. Two businesses means two installations, and we would price that as a conversation rather than pretend one licence stretches.

No, and it does not try to. Your DMS holds stock, deals and invoicing. Bosun holds the work people agreed to do about them - the callbacks, the recon steps, the parts chasing. Those are the things a DMS has never tracked well, which is why they live in a group chat.

They already use WhatsApp all day, and this is WhatsApp. That is the entire bet, and it is why there is nothing to install or log in to. The one thing it asks of them is a message at the end of the day.

Departments can be per site, which works for two or three locations under one owner. Genuinely separate businesses want separate installations - tell us the shape and we will say what fits.

Yes - a starter set of approval limits, recon steps and service processes, editable and meant to be replaced by your real documents. It exists so a demo on day one can actually answer a question.

No. Your existing system holds the job cards, parts and invoicing. Bosun holds what people said they would do and whether they did it - which is the part that currently lives on a whiteboard and in messages.

Then they do not have to type. They can hold the button and say it - "rotors came in, KX19 done, Transit still stuck on the turbo" - and that voice note is transcribed and handled exactly like a typed message. It repeats back what it heard, so a mis-heard registration is caught before it becomes a job. Typing works the same way and takes about nine seconds. See <a href="/features/voice-notes/">voice notes</a>.

Give them the manager role and yes - they can see and assign across the workshop. Technicians see and change only their own work.

No. Treat it as the thing that makes sure the paperwork gets done and gives you a searchable record of who said what. If your inspector requires a specific format, keep using it.

Not mid-service, and they should not. The realistic pattern is a message before service and one at close-down - which is exactly when the checks happen anyway.

Yes. Describe the check once - what it is, how often, what time, who it belongs to - and the job appears on its own every day, or on the weekdays, or once a week or month. From the moment it appears it is an ordinary job: chased if it goes past its time, and visible on the owner's briefing if nobody confirms it. See <a href="/features/recurring-work/">recurring work</a>.

Compliance is a property of how you use a system, not a badge software carries, so the honest answer is: it can be part of a compliant setup, and it is not compliant on its own. It runs on your server, we hold no copy, and it needs no patient identifier. What you must assess is your AI provider's terms and your own policy on staff messaging. We will give your data protection officer straight answers, including where the answer is unhelpful.

No. Your PMS holds appointments, records and billing. Bosun holds the follow-up work around them - the chasing and the calling that currently depends on somebody remembering.

Yes, and many practices should. Put reception and administrative staff on it, leave clinical notes where they belong, and you get the chasing handled without touching clinical systems at all.

In one file on the server you chose, in the country you chose. That is the whole answer, which is the point of self-hosting.

Not as things stand - it refuses numbers that are not on your staff roster, by design, so it is not a contractor portal. Your staff record what a contractor said, which is where the chasing needs to happen anyway.

No. Your existing system holds tenancies, rent and accounts. Bosun holds the maintenance work and the chasing, which is the part that currently lives in messages.

Yes, if the address is in the job titles - and it will be, because staff naturally type it. Search the address in the dashboard and every job for it comes back with its history.

No. It works on whatever phone has WhatsApp, including their own - which is why there is nothing to roll out. Whether you want work on personal phones is your policy call, not a technical limit.

WhatsApp queues the message and sends it when they have signal again, exactly as it does for their personal messages. Nothing is lost, and the assistant processes it whenever it arrives.

Yes. A photo is kept and attached to the job that message created, so the picture of the finished work sits on the job about it. Voice notes are transcribed and handled like any typed message, which is usually what a van actually sends. What it does not do is interpret the photo; a person still looks at it. If photo capture matters to you, say so - it is a reasonable thing to add and we would rather know it is wanted.

Comfortably fifty on the smallest server, and the price does not change at any number. Beyond that, tell us the shape and we will be honest about what needs testing first.

No. There is no gate, no download form and no drip sequence. If a guide is useful you will remember where you read it, and that is a better outcome than an unsubscribed address.

They mention it, near the end, where it is genuinely relevant. The first half of each one is advice that works with a notebook and no software, because a guide that sells from paragraph two is worth nobody's time.

No. It is unprofessional to lose a customer's callback, and that happens on expensive platforms too. Use the channel your team will actually use, and put the discipline into how you use it.

It is a policy question with no universal answer. What matters practically is that the tool works on whatever phone they have, so you are choosing freely rather than being forced by software.

Naming people works immediately. Daily updates take about three weeks to become a habit, and they fail in week two if nobody visibly reads them.

Then start with the smallest one - naming a person in every request - and do only that for a month. Four changes at once is how none of them happen.

Make them expected rather than punished. "Everybody sends one" is a norm and it works; "you will be disciplined for missing one" produces three words of nothing every day at 17:59.

Then two words and a phone call is fine, and you write it down. The point is the information reaching a place that can be checked, not the medium.

For your process, yes - it is better than nothing. Be aware that a voice note is not searchable later, so somebody has to turn it into text if it matters.

Ask about blockers rather than output. "What are you stuck on?" is help; "what did you do today?" is inspection. The information you get back is almost identical.

Five. The five questions you are asked most often, one page each. Anything more on day one and you will not finish.

Whatever you will actually keep updated. Plain text or Markdown files are easiest for software to read reliably; a shared document is fine if that is what you will maintain.

Anything touching employment terms, health and safety or regulated activity, yes. "How long should recon take?" does not need a lawyer, and treating it as though it does is why handbooks never get written.

You have found something valuable. A contradiction between documents means two people have been operating differently, and now you can decide which is right instead of finding out from a customer.

Two weeks of trialling and one afternoon of arithmetic. Longer than that and you are avoiding a decision rather than researching one.

Involve the person who will resist it most. They will tell you in five minutes what a three-month rollout would have taught you expensively.

Yes, genuinely. Set up a free board for a week and see whether your team keeps it current. If they do, you have saved yourself a purchase - and if they do not, you have learned the most important thing about your team.

You will have spent some money and learned what your team actually does. Most businesses buy twice. The point of the twelve questions is to make the first one the right one.

If your team is under five people, none of them - your group chat is fine. If you need rotas, a time clock and payroll, Connecteam. If your team will genuinely open an app, Trello is cheap and good. If the problem is that they will not, and you want to own the software outright, Bosun.

Because these are the ones that actually come up on buyers' shortlists next to us. A page comparing us to thirty tools nobody considers would rank well and help nobody.

They were read off each vendor's pricing page on the date printed at the bottom of each comparison. Prices change without notice - check the source links before you rely on a figure.

You get it, and you apply it when it suits you. The first year of support is included; after that, updates and help are a paid renewal you can take or leave - and declining it does not switch anything off.

There are running costs, and they are not hidden: hosting at roughly $12 to $25 a month, your own AI key at roughly $5 to $15, and WhatsApp's own fee on messages sent outside the 24-hour window. All paid to those providers in your own name, never to us.

It is, because setup is where the work is and support is bounded by there being one installation per buyer rather than a shared platform to keep online. If it ever stops being sustainable, the price changes for new buyers - never retroactively for you.

Ask. We would rather split the fee over a few months than lose a buyer who wants the product - but it is a payment arrangement, not a subscription, and it ends.

No, and it refuses them on purpose. A number that is not on your staff roster is rejected before anything is stored. Mixing staff task tracking with customer support in one assistant would make both worse.

It has ticketing and labels built around conversations, which some teams use as a loose task list. What it does not do is read a plain-language plan, create jobs from it, chase them and report on throughput - check their documentation rather than taking our word for it.

Bosun, from about the eighth month onwards at fifteen seats - and the gap widens every month and every hire. Periskope is cheaper on day one, which is a real advantage if cash today matters more than cost over three years. Run your own figures in the <a href="/cost-calculator/">cost calculator</a>.

It depends on exactly what it does, and we are not going to make a legal claim about a specific product. What we will say is that automating WhatsApp Web sits outside the official Business API, that Meta acts against unofficial automation, and that the number carrying your business is a bad thing to gamble. Ask the vendor whether they use the official API and judge the answer.

Technically yes, since they touch different numbers - Bosun uses a Business API number, extensions work on personal WhatsApp Web. Whether you want to pay two bills is a separate question.

It is not - $1,999 is real money on day one, and a per-seat tool is cheaper in month one at every team size. The figures above are three-year totals, and we would rather show the crossover than pretend there is not one.

Ask whether your staff will reliably open an app. If they will, Connecteam is cheaper and does more - buy it. If you have already watched an app die on your team, that is what Bosun is for, and it is worth the difference.

No. We are more expensive over three years and the page says so. What we will do is tell you before you buy if we think you are buying the wrong thing.

You can wire a message into a card with an automation tool, and some people do. What that does not give you is understanding plain language, matching an update to the right existing job, chasing and escalating, or answering from your procedures - which is the majority of the work here. If a simple bridge solves your problem, build it and keep your money.

No. It has a list ordered by what needs deciding, filters and a full history per job. If a visual board is what you want, that is a genuine reason to choose Trello.

That is the most common situation we hear. Before switching, try one thing: ask people why. If the answer is "I forget to open it", Bosun addresses exactly that. If it is "the work is not organised", new software will not fix it.

No, and it does not go in it. The assistant works in one-to-one chats with each person; your group carries on for everything it is already good at. Nothing is taken away.

Probably not yet. At four people a group chat genuinely works, and we would rather tell you that than sell you something you do not need. Come back when chasing has become somebody's morning.

Yes, and you should. The banter, the shift swaps, the photographs - none of that belongs in a job list.

Nothing. <a href="/book-a-demo/">Book a demo</a> and you can send it real messages from your own business and see what comes back, including the live running cost.

The product is new. It was built against one dealership's real operations, tested extensively - 716 automated checks including real-browser tests of every screen - and is now being sold. If you would rather buy software with ten years of history, that is a legitimate preference and we are not it yet.

Few, and we are not going to inflate the number. Early buyers get more of the founder's attention than later ones will, which is the honest trade for going first.

Setup slots get booked further out, and we will tell you the real date rather than a hopeful one. Support for existing installations comes before new sales.

Yes, and you should. <a href="/book-a-demo/">Book a demo</a> and you will speak to the person who built it. Bring the twelve questions from the <a href="/guides/team-task-software-buyers-guide/">buyer's guide</a> - they were written to be asked of us too.

Compliance is a property of a whole setup rather than a badge software carries. What we can tell you is what matters for your assessment: your data is on a server you chose, in a country you chose; we are not a processor because we receive none of it; and the one external flow is the message text going to your own AI provider under your agreement with them. Your data protection officer will get straight answers from us, including where the answer is unhelpful.

Nobody, unless you give somebody access - which some buyers do during setup and then revoke. There is no standing access, no support backdoor and no account on our side.

WhatsApp messages are encrypted in transit, and Bosun uses Meta's official Business API. It is still a consumer messaging app on staff phones, so the sensible discipline is to keep identifiers and confidential detail out of messages. For a clinic, that discipline is the actual safeguard - see the <a href="/solutions/clinic-task-management/">clinics page</a>.

No. It needs to reach WhatsApp and your AI provider. An air-gapped installation is not possible with the current design, and we would rather say so than talk around it.

Less per installation than a single purchase, and the actual figure depends on how many and how much of the setup you do yourself. It is a short conversation, not a negotiation.

Yes. The product carries no reference to us, and if you handle setup and first-line support your clients never need to know we exist.

Ask. For a referral that turns into a sale, we would rather agree something than have you do it for nothing - but it needs to be written down, like everything else here.

No. Install it for one client, see how it goes, and come back if it worked. A minimum would only push you into selling it to somebody it does not suit.

Because the comparison and pricing pages make factual claims about other companies, and a claim with no stated standard behind it is just an assertion. It also gives you something concrete to hold us to.

The writing is drafted and edited by the founder, with AI used as a tool in the process - the same as most software is now written. What matters is that a named person is accountable for every claim on it, and every figure was checked by that person against the product or a linked source.

Because there are not yet customers who have agreed to be quoted. Inventing them would be the easiest thing on this site to fake and the fastest way to deserve no trust. When real ones exist, they will appear with real names.

No. It runs entirely in your browser, there is no form to submit, and nothing is stored or sent. Reload the page and it forgets everything.

Because they are a worked example and we picked a realistic one - fifteen people against a $20 seat. The section above tells you exactly which three numbers to change to make it argue the other way, which is a fairer thing to publish than a neutral-looking default.

Because converting a competitor's published dollar price into another currency would mean inventing an exchange rate and presenting it as a fact. Type your own figures in your own currency and the comparison still holds - only the symbol is wrong.

Yes, measured on live traffic through a running installation with prompt caching enabled. Without caching it is roughly five times that, which is worth knowing if you are comparing against a tool that does not cache.