Auto-reply: answer the first message before you get to it
Somebody writes to you at 19:40. Nobody is at the console. They see their own message and nothing else, and have no way to tell whether it arrived, whether anyone will read it, or when.
The auto-reply is your workspace answering that in your own words, before any teammate gets there. It is on by default, with text we ship, so it works in a workspace that never opens the settings form.
It is not an AI. It says exactly the sentence you wrote — or ours, if you wrote none.
Where it lives
Widget → 4. Auto-reply. A switch and two text boxes:
| Field | What it is for |
|---|---|
| Switch | Whether the workspace answers automatically at all |
| During working hours | What a customer reads when you are open |
| Outside working hours | What they read when you are not |
Up to 500 characters each. Leave a box empty and we use our own sentence, in your workspace's language — an empty box means "use the default", never "say nothing". Saying nothing is what the switch is for, and switching it off keeps your texts, so turning it back on does not mean typing them again.
Which of the two a customer gets
Your working hours decide — the same schedule you set in 3. Working hours, in the time zone you chose there. Inside them, the first text; outside, the second.
A workspace with no schedule is always open, so it always sends the first text. That is the same rule the waiting-visitor alert uses, so the two can never disagree about whether you were around.
It is not an answer, and nothing pretends it is
This is the part worth reading twice.
A conversation the workspace greeted automatically still says Needs reply. It stays in the awaiting-reply filter, still counts in the sidebar badge, and the waiting-visitor alert still fires for it on its usual schedule. The customer was promised an answer, not given one.
In the thread your team sees it marked Auto-reply, so nobody reads it as a colleague's words — and it carries no edit pencil, because nobody typed it and there is no wording of anyone's to correct.
In the widget
The customer sends their message, sees the usual "Message sent" line, and a moment later the answer appears in the panel — the same way an operator's reply arrives, because that is exactly the route it takes. It is a message in the conversation, not a banner: it stays in the thread, and they can still see it when they come back tomorrow.
Every new conversation gets one, not only the first one a browser ever starts. Somebody writing about a new problem next week is asking a question nobody has read yet, and is owed the same promise.
In email
A letter to your workspace address is answered too, and the answer really is a letter: it arrives in the thread the customer just started, from your workspace address, signed with your workspace name — no operator name, because no operator was involved.
Three differences from the widget, all of them about not answering machines:
- Only the letter that opens a conversation is answered. A reply inside a thread your team is already handling gets nothing — by then the promise is over.
- At most one automatic answer per correspondent per day. A customer sending three separate requests in an afternoon is a real customer, but this bound is what stops a badly-behaved mail robot from trading letters with us all night.
- The answer is marked as machine-written (
Auto-Submitted, the mail standard for this), so a well-behaved autoresponder on the other side stays quiet. In the other direction we do the same: out-of-office replies, mailing lists and bounce notices never become conversations at all, so they are never answered.
Turning it off
Switch it off and the workspace answers with nothing at all — no message in the thread, no letter. The customer still sees the "Message sent" line in the widget, and their letter still reaches your inbox exactly as before.
The texts stay saved.
What to write in it
Short, and about waiting rather than about you. Two useful habits:
- Make the promise you can keep. "We reply within a few minutes" is worth a lot when it is true and costs you a customer when it is not. If the honest answer is "by tomorrow morning", write that.
- Use the closed text for the concrete hour, not for an apology. "We're back at 9:00 on Monday" tells somebody whether to wait or to phone; "sorry, we are currently unavailable" tells them nothing they did not already know.