Cold email with a reason, from your own mailbox
Each email names something dated that changed at the company. You approve it, it sends from your mailbox inside the daily cap, and replies are sorted for you.
Each test runs
14 days
The short version
The AI agents pick the companies you watch where something dated changed, find the right person, and draft an email that names that evidence. You approve every email, it sends from your own mailbox inside daily caps and warm-up limits, and the test is judged on the positive reply rate against bars fixed before the first send.
Read against the channel's code on 2026-10-13
How one test runs
Success bar
At least 100 sent, 2% or more positive replies
Stop bar
At least 100 sent, half a percent or fewer positive replies
Where it gets more
At least 100 sent, 4% or more positive replies
What the agents do
- Picks the watched companies with the most dated evidence, and finds the person who now owns the problem.
- Drafts each email around that evidence, with its source and date beside it for you.
- Sends what you approved from your mailbox, inside your daily cap and the warm-up limits.
- Sorts the replies and drafts the follow-ups.
The AI agents on this channel
What you do
- Connect the mailbox the emails go from, Google or Microsoft.
- Watch at least one company, or accept the ones Discovery suggests.
- Approve every email before it sends, and answer the positive replies.
The first two weeks, as planned
Example for a fictional company: Harbourline, a fictional B2B software company
| Day 1 | Pick the accounts with the most dated evidence and find the right person at each | AI agents |
|---|---|---|
| Day 2 | Draft the first ten emails, each naming its evidence | AI agents |
| Day 2 | Approve the first ten emails | You20 min |
| Day 3 | Send inside your daily cap and the warm-up limits | AI agents |
| Day 7 | Classify replies and draft follow-ups | AI agents |
| Day 8 | Approve follow-ups and answer positive replies | You30 min |
| Day 14 | Measure the positive reply rate against the criteria | AI agents |
Before it starts
- Connect the mailbox the emails go from
- Watch at least one company
What your weekly report shows
- Sends, replies and positive replies against the success and stop bars.
- Bounces and complaints, because they decide whether the mailbox stays healthy.
- Meetings and deals that came from the channel, when you track deals.
What it counts
- Items drafted
- Sends
- Replies
- Positive replies
- Reply rate
- Positive reply rate
- Meetings booked
- Bounces
- Bounce rate
- Spam complaints
- Negative replies
- Drafts you rejected
- Meetings held
- Qualified deals
- Pipeline value
- Deals won
- Revenue
What it will never do
- Every send passes one check first: suppression, excluded domains, per-domain caps, mailbox health, your approval and the budget.
- Emails send from your own mailbox, never from ours.
- An email never mentions an event without a stored source and date behind it.
- Sending needs a paid plan. The Free plan drafts the emails and cannot send.
Questions people ask
Something specific and dated about the company: a hire, a launch, a change in the team. The evidence sits beside the draft with its source and date, so you can check it before you approve. A sentence with no stored evidence behind it is refused.
No. You approve every email before it sends, and each send passes the check on suppression, caps, mailbox health and budget at the moment it goes out.
It sends inside a daily cap and the warm-up limits, and counts bounces and complaints against the stop bar. A channel that bounces or draws complaints stops before it costs you the mailbox.
Only companies you watch or accepted from Discovery, and only where something dated changed. Domains you excluded are never written to.
Cold email with a reason, from your own mailbox
Free. No card. Ready while you wait.