Only things that are live. No roadmap, no “coming soon”. If it is on this page, your account already has it.
Thousands of catalogued products sat with no contact address and nothing was draining that queue, so they were invisible to matching no matter how well they fit. A daily sweep now works through it and every site ends up with an address or a reason it has none.
Addresses that no longer accept mail used to sit in the directory until they bounced. A daily pass now checks format and whether the domain still takes mail at all, and clears what does not, instead of slowing your sending down to hide the problem.
When someone answers, every pending send to that address stops the same minute and the lead is marked as replied. Before this, a person who had already written back could still receive the next proof.
A lead skipped because your endpoint found nothing that day is not a dead lead: your product changes and so does theirs. Once the directory has grown enough for the picture to be different, those leads return to the queue for one more run.
A bounce used to disable the whole prospect. Now it only burns that address: the company stays in the directory and goes back to the queue as soon as a different, working address is found for it.
Replies are now tied to the exact finding that triggered them, not just the campaign. The sweep view shows which evidence types (probe failures, doc mismatches, timing data, actually start conversations, per account.
A spam complaint now suppresses that address across every account on the platform within the hour, not just for the sender who triggered it. Previously suppression was per-account with a daily global sync. The sending policy page was updated to match.
Findings that score below the confidence threshold but above the discard line now land in a review state instead of being dropped silently. You see the finding, the reason it was flagged, and a one-click send-or-discard. Nothing in the review queue sends on its own.
Daily sends are now spread across a configurable window instead of going out in one morning burst. Default window is 08:00 to 17:00 in the recipient’s likely timezone, minimum spacing five minutes between sends from the same account.
Soft bounces now retry on a fixed schedule of 1h, 6h and 24h, and then park the address for 30 days instead of retrying indefinitely. Permanent bounces still kill an address on the first response. Behaviour is documented on the sending policy page.
If two customers match the same launch on the same morning, both proofs are built but the second sender is queued to the following day and shown a notice. One founder should not get three near-identical probe emails in an hour; that is how the whole approach dies.
The morning sweep now covers nine launch directories plus Show HN and the Product Hunt feed, up from five sources at launch. Typical yield is around 200 new products a day to match against.
First production run: sweep, ICP matching, proof building against the customer’s own system, paced sending with a daily cap per account, and full bounce and unsubscribe handling from day one. If a run finds nothing worth saying, nothing is sent, and that rule shipped before the signup form did.