Product

Email sequences that land

The Harp team ·

An AI agent can write hundreds of emails in a session. Sending them that fast is how domains get burned. Harp’s job is to let your agent decide what to send, while Harp decides when, from which mailbox and how many.

How Harp sends

  • From your own Google Workspace mailbox through the Gmail API, not shared sending servers.
  • A 28-day warm-up for every new mailbox, from 10 cold emails a day in the first week to 40 in the fourth, never above the mailbox’s own daily limit.
  • A random 2 to 6 minutes between sends from each mailbox.
  • Delivery in the recipient’s working hours, in their timezone, when it is known.
  • A plain-text version with every email, alongside the HTML.
  • Open and click tracking off by default; a workspace can turn it on.
  • A one-click unsubscribe header with a signed link per recipient.

Sequences that know when to stop

A reply, bounce or unsubscribe stops that person’s sequence immediately, and the suppression list is checked before every send. Sequences can mix email with LinkedIn steps, waits and conditions.

Drafts you can review

Emails render from your templates, with your agent’s research answers as variables. If an answer is missing, the email is held rather than sent with a gap. A campaign can require copy review, so your agent reads every rendered email before it goes out, and copy rules you save apply to every draft after.

Questions

How many emails can one mailbox send?

Each mailbox has its own daily limit, lowered while it warms up. With 2 to 6 minutes between sends, one mailbox sends 10 to 30 emails an hour at most.

Do I need a separate domain for cold email?

We recommend one, with SPF, DKIM and DMARC set up, so your main domain’s mail is never affected.

Run this with Harp

The MCP harness for pipeline generation. We set up every customer ourselves.

Talk to sales