WhatsApp puts a business number your company owns inside Quanty. A message a customer sends to that number arrives as a row in a table, an automation can answer it, and the same AI Secretary that answers your phone can hold the whole conversation in writing. Once a message is a row it is ordinary Quanty data: filter it, put it on a board, read it with AI columns and ask chat about it.
A number is connected together with us, not by you alone
Linking a WhatsApp business number is not a two-click, self-serve step, and that is Meta's requirement rather than a missing feature in Quanty. Meta verifies the business itself, on company documents, in a review that takes days rather than minutes, and the linking flow signs into the Meta business account that owns the number. So we connect it together with you, one company at a time.
Connecting a number
The number is yours and stays yours. Quanty does not sell numbers and cannot create one for you, and the linking step happens in the AWS console rather than in the app, because that flow signs into Meta and needs someone who can act for the business.
What you need first
- A number your company owns, in international format with the country code, on which you can receive a verification code.
- A number that is not already in use in WhatsApp. It must not be signed in to the WhatsApp app or to the WhatsApp Business app. If it is, remove it there first: a number belongs either to one of those apps or to the business platform, never to both at once.
- A business number, not a personal one. This is the number customers will see as your company.
- Someone who can act for the business at Meta, because the linking flow signs into the Meta business account.
How the linking goes
- In AWS End User Messaging Social, the "Add WhatsApp business account" flow is started and signed in with the Meta business account that owns the number.
- Meta verifies the business. That is their review, on their documents and at their pace, and nobody can submit it for you from inside an application.
- Once the number is registered, it is added in Quanty with the identifiers AWS shows for it: the business account id, the AWS phone number id, Meta's own id for the same number, and the number as a customer sees it.
- The number appears on the WhatsApp page with its status, its recent messages and a test send, so you can watch one message land on a phone before anybody else gets the number.
Steps 1 and 2 are the ones that take days. Step 3 takes a minute.
Living with a connected number
- A number can be paused and resumed. Paused, it sends nothing and receives nothing, and the automations that use it stop doing anything.
- Removing a number stops Quanty sending and receiving through it. Nothing changes in AWS or at Meta, and the messages already received stay where they are.
- One number can be used in one organization only.
- Only an administrator of the organization can add, change or remove a number.
The display name
The name customers see next to your messages belongs to Meta, not to Quanty. Meta reviews it, and changing it afterwards means a support ticket with them. Decide what you want to be called before the review rather than after it.
When a WhatsApp message arrives
Triggers & automations gains a trigger type: A WhatsApp message arrives. It runs every time a customer sends a message to one of your numbers. Pick the number it listens on, or leave it on Any number.
A trigger only ever sees what comes next: messages that arrived before you turned the automation on never run it.
Filters
Under Only when, narrow it down to the messages you care about:
- From this number: leave it empty for any sender, or give the numbers in international format with the country code.
- The text contains: a word or a phrase from the body of the message.
- The message has an attachment: only messages carrying a photo or a document.
Separate several values with commas and a message counts when any one of them fits. An empty box does not filter at all.
Saving the message as a row
Create a row for the message writes one row per message into the table the automation runs on, and Put the message in these columns decides where each part of it lands:
| Field | What lands in the column |
|---|---|
| Sender's number | The customer's number, in international format. |
| Sender's name | The name on their WhatsApp profile, as they set it themselves. |
| Text | What they wrote. |
| Received at | When the message arrived. |
| Attachments | A file column: the photo or the document they sent. |
Every field can also be set to Do not save it.
A profile name is not an identity
The sender's name comes from the customer's own WhatsApp profile. Anyone can put anything there, so treat it as a label and not as identification.
What runs after that
Once the row exists, every action an automation already has applies to it: run the AI columns over the message, notify people, send an email, create or update a record, call a webhook, and send a WhatsApp message back.
Send a WhatsApp message
The action Send a WhatsApp message sends from one of your numbers.
Who it goes to
Three recipient sources, one per action:
- These numbers: typed in, one per line, in international format with the country code.
- Numbers from these columns: the value of the column is read per row and used as the recipient, so one rule can message everyone a view contains.
- Reply to whoever wrote: available only in an automation that a WhatsApp message started. The answer goes back to that sender, on the number they wrote to.
The message
The text is a template. Put a column name in curly braces and that row's value is dropped in: {{Order number}}.
On a run that a WhatsApp message started, the message itself is available too:
{{whatsapp.from}} (the customer's number), {{whatsapp.from_name}} (their profile name), {{whatsapp.body}} (what they wrote), {{whatsapp.received_at}}, {{whatsapp.media}} (the attachments) and {{whatsapp.number}} (your own business number, the one the message arrived on).
The builder lists them next to your columns, so nothing has to be typed from memory.
What to send is either Free text or An approved template. Which of the two you may use is not your decision, and that is the next section.
The 24-hour rule
This is WhatsApp's rule and not Quanty's, and it is the one thing that surprises everybody.
A customer who writes to your number opens a 24-hour window, measured from their last message. Inside that window you answer in free text, as often as the conversation needs. Outside it, WhatsApp accepts only a template Meta approved in advance, never free text.
Nothing can warn you in advance that a particular send will be refused, because only WhatsApp knows when the customer last wrote. Free text outside the window comes back refused; it does not wait in a queue until the customer writes again.
Rules the window makes easy
Anything that answers. The customer writes, the automation runs, the reply goes out in the same minute:
- a reply to whoever wrote in, with a confirmation, a reference number or a value read out of the row;
- an AI column reads the message and the automation sends back what it concluded;
- a secretary conversation, which is nothing but a reply to each message the customer sends.
Rules that need a template
Anything you start, and anything that waits. If hours or days have passed since the customer last wrote, or they never wrote at all, only an approved template goes through:
- a status change on a row that the customer is told about the next day;
- a scheduled reminder;
- a message to the numbers held in a column, sent to people who have not written to you.
You register the template with Meta in the WhatsApp Manager. Once Meta approves it, the action takes its Template name exactly as Meta has it and the Template language, the language code of the approved version, for example pl or en. Approval is Meta's to give, so have the template before you build the rule around it.
The AI Secretary on WhatsApp
A number can point at an AI Secretary you already run, and the secretary then answers messages in writing. There is no second secretary to build: the persona, the knowledge about your organization, the rules it cannot break and the list of fields to collect are the ones you configured for the phone.
When the conversation finishes it writes one row into the same table a phone call writes into, onto the same columns, and the automations on that table fire once, exactly as they do for any other new row.
It stops on purpose. Once it has asked enough, it sends a hand-over line and closes the conversation for a person to pick up instead of talking in circles. A conversation nobody continues is closed by itself and written away rather than left open, so a customer who went quiet still ends up as a row somebody can answer.
The first message the secretary sends carries the same notice a caller hears: the person is told that they are writing to an AI assistant. That notice cannot be edited or switched off.
What it costs
WhatsApp is part of the Gold plan and up, the same tier as the AI Secretary and KSeF, and for the same reason: a WhatsApp conversation costs money outside your credit balance.
Three separate things are paid for:
- Meta charges for the conversation, on its own price list, which depends on the country and on the kind of conversation.
- AWS charges for carrying each message.
- Quanty credits pay only for the AI work: an AI column reading a message, or one turn of a secretary conversation. A message with no AI in it costs no credits.
Quanty does not meter the message fees for you. They arrive on the Meta and AWS bills rather than in your credit usage. See Billing & credits.
Where the data goes
A WhatsApp message passes through Meta, which runs WhatsApp and holds the conversation on its own terms, and through AWS End User Messaging Social in Frankfurt (eu-central-1), which is how Quanty sends and receives. What Quanty then keeps, the message, the row and any attachment, sits in AWS Frankfurt with the rest of your workspace.
A secretary conversation on WhatsApp does not take the path a phone call takes. It is text, it never reaches the voice provider, and it runs on the same AI models as the rest of the app. Security & MFA says where those models run.
Message bodies and full phone numbers are never written to Quanty's logs. The conversation itself is stored, because storing it is the feature.
The current list of the companies that process data on our behalf is at quanty.ai/subprocessors.
Related pages
- Triggers & automations: the trigger and the action this guide describes.
- AI Secretary: the persona, the fields and the table a conversation writes into.
- Integrations: the hub the other connectors sit in.
- Security & MFA: where your data is processed and who can see it.