One inbox for every reply, sorted by what it means

The problem with many mailboxes

Cold email at any real volume means many sending mailboxes. Replies scatter across all of them, and the two ways teams cope — forwarding everything to one Gmail, or logging in one by one — both lose the thing that matters: who said what, to which message, and whether anyone has answered yet. Meanwhile the sequence keeps sending follow-ups to someone who already replied.

What MailFleet does instead

Every mailbox is polled every 30 seconds. Replies from all connected senders land in one threaded queue. A reply reopens its thread with an unread badge so it resurfaces; the nav shows a live count.

The sequence stops the instant a reply arrives — before anyone reads it. By default everyone else at that company is paused too, so a "please stop" from one person doesn't get followed by a cheerful step three to their colleague.

Each reply is tagged with its intent: Interested, Meeting request, Question, Not now, Not interested, OOO, Wrong person, Referral. Rules and a language model read the reply; the tag shows on the thread, filters the queue, and fires the reply.tagged webhook. You choose per campaign which tags may be applied.

Out-of-office replies are handled properly. If the auto-reply gives a return date, the next step and any open task wait until that day rather than firing into an empty inbox or ending the sequence. Without a date, the sequence carries on. A real reply from the person ends the hold.

Bounces never masquerade as replies. Delivery failures go to their own Bounced view; only permanent failures (no such user) count as bounces, and they stop the lead's sequence.

Working the queue

• Folders for the way you work: Inbox, Unread, Important, Snoozed, Archived, Untracked and Bounced, plus Interested and Needs an owner views. Search every message; filter by category, email status, campaign, email account, team member, campaign tag, client or reply date.
• Each lead as one timeline: the emails they were sent with opens and clicks, their replies, your answers, notes, follow-up tasks and forwards.
• Reply inline from the address the person wrote to, staying on your warmed mailbox and inside the same thread, with up to 10 attachments of 5 MB each. The sequence stays stopped. Forward a reply to a colleague from the same mailbox.
• Mark lead as Interested, Meeting booked, Not interested and the rest — the auto-tagger's tags, so subsequences fire just as they would automatically. Star, snooze, archive, assign, add a follow-up task or a private note.
• Call Lead from the thread, Push To Campaign, Replace Lead when you reached the wrong person, pause a lead, or block their whole domain — with two-key keyboard shortcuts.
• Log a reply by hand (a phone call, a reply that came to a personal inbox) — it stops the sequence exactly like a real inbound.

Subsequences: act on the tag

A subsequence is a short follow-up sequence that starts the moment a reply gets a chosen tag. "Meeting request" can send the calendar link straight away, from the mailbox the person replied to, in the same thread. "Referral" can thank them and ask for the introduction. You write the steps in the same builder as the main sequence.

LinkedIn replies in the same queue

Every connected LinkedIn account's inbox is read every 20 minutes. Messages from prospects are filed as replies tagged "LinkedIn", matched to the lead by profile URL or name, and treated like email replies: sequences stop across both channels, auto-tagging and subsequences run, webhooks fire.

Getting replies to where your team already works

• Reply forwarding — a default address, plus rules so the replies certain mailboxes receive go to a specific inbox. Replying from there answers the lead from that inbox; replying inside MailFleet keeps the conversation on the warmed mailbox.
• Webhooks on every plan — reply.received, reply.interested, reply.tagged, email.bounced, lead.unsubscribed and three more, signed with HMAC-SHA256, retried, with a delivery log.
• API (Pro and Scale) to pull replies, leads and campaign stats.
• Claude connector — list replies or read stats from a chat.

Measuring it

The campaign page shows replies and interested replies as counts and rates; the Overview shows how many people wrote back and starts you there. A reply-health check in the operator console tests any mailbox's IMAP connection on demand, because a mailbox that can send but not receive is a queue with a hole in it.

FAQ

Does the sequence stop as soon as someone replies?

Yes — automatically, whether or not you have read the reply, and by default the whole company is paused too.

Can I reply from MailFleet, or do I have to open the mailbox?

From MailFleet, in the thread, from the exact address the person wrote to. Attachments are supported.

How are replies classified?

Rules plus a language model assign one tag: Interested, Meeting request, Question, Not now, Not interested, OOO, Wrong person or Referral. You can add your own tags and choose per campaign which automatic tags are allowed.

What about out-of-office auto-replies?

If the auto-reply states a return date, follow-ups are held until that day. If not, the sequence continues. A real reply from the person ends the hold.

Do LinkedIn messages come into the same inbox?

Yes. Each connected LinkedIn account's inbox is read every 20 minutes and prospect messages are filed as replies tagged "LinkedIn".