If you rejected a draft and updated your instructions, your campaign no longer gets stuck waiting behind it. It now generates a new draft against the updated guidance and resumes sending on its own.
What changed
Previously, rejecting a draft and then improving your campaign instructions, for example correcting something in the Learning tab after seeing a draft that missed the mark, didn't reliably trigger a new draft. Some campaigns stayed stuck behind the rejected draft, with sending paused for that contact and no new draft generated to replace it. From your side, it looked like the campaign had simply stopped, with no clear reason why and no obvious next step to unstick it.
That's fixed. When you reject a draft and update your instructions, the campaign now generates a new draft against the updated guidance automatically, without you needing to trigger it manually or restart anything. Once that new draft is approved, either by you or by whatever review step you have configured, sending resumes cleanly for that contact instead of staying paused.
How to use it
Reject a draft the way you normally would when it doesn't match what you want, whether that's a tone problem, a missing detail, or an instruction the AI didn't follow correctly. Go to the Learning tab or your campaign instructions and update the specific guidance that caused the issue, the same way you'd fix it for any other draft.
A new draft generates against the updated instructions. Review it as you would any draft: check that it reflects the correction you made, and approve it once it looks right. Sending picks back up automatically for that contact from there; you don't need to manually re-trigger the campaign or check that it's moving again.
If you have a campaign that was stuck from before this fix went live, check its status now. It should show either a fresh draft waiting for your review or sending already resumed, rather than sitting paused with no activity.
Why it matters
Rejecting a bad draft is only useful if the campaign actually acts on the feedback behind that rejection. A campaign that stays silently paused after you've already done the work of identifying what was wrong and fixing your instructions wastes both the correction and the time you spent making it, and worse, it means a contact who should be getting outreach simply isn't, with no indication that anything has gone wrong until you happen to notice.
Redrafting and resuming automatically closes that gap. The fix you make in the Learning tab actually reaches your next draft for that contact, instead of sitting there unused while the campaign waits on a draft you already rejected. That matters most for campaigns with a lot of contacts moving through at once, where a stuck contact here and there is easy to miss until it's added up to a meaningful gap in your outreach.
This was also inconsistent in a way that made it hard to diagnose from the outside. Some campaigns recovered fine after a rejection and an instruction update, which meant the ones that didn't looked less like a systemic issue and more like something specific to that campaign or that contact, making it harder to know when a stuck campaign was actually a bug versus something else worth investigating on its own.
Iterating on a new campaign's voice
If you regularly reject drafts as part of refining a new campaign's instructions, this now behaves the way you'd expect that loop to work: reject, correct the instruction, get a new draft that reflects the correction, approve, and the campaign keeps moving. You can iterate on instructions this way with confidence that a rejection doesn't risk quietly stalling the campaign behind it.
For campaigns with many contacts, it's still worth spot-checking a few contacts after a broad instruction change, not because the redraft-and-resume behavior is unreliable, but because a broad change in Teach or the Learning tab can affect drafts in ways worth reviewing before they reach a large number of prospects at once.
What this means going forward
Beyond the specific stuck-campaign symptom, this fix restores a basic expectation about how feedback is supposed to work in the product: correcting an instruction should visibly change future output, not require you to separately verify that the correction actually took effect. That expectation matters most while you're still tuning a new campaign's voice, which is exactly the period where you're rejecting drafts and adjusting instructions most frequently.
It also removes a reason to hesitate before rejecting a draft in the first place. If rejecting carried a real risk of quietly stalling the campaign, the safer-feeling option was sometimes to approve a draft that wasn't quite right rather than risk breaking the flow of sends. With reject-and-regenerate working reliably, there's no longer a reason to make that trade-off; rejecting a draft that misses the mark should always be the better choice than sending it anyway.
This applies whether you're tuning a brand-new campaign or making a correction to one that's been running for weeks. A campaign that's been sending reliably for a while and suddenly produces a draft that's off is just as likely to benefit from this fix as a new campaign still finding its voice; reject it, fix the instruction, and expect the campaign to pick right back up rather than needing manual attention to get it moving again.
If you've been in the habit of periodically checking older campaigns for stalled activity as a precaution, that check is worth doing once more now to confirm anything affected has cleared, after which it shouldn't be a recurring concern tied to this particular pattern. The behavior you should expect from here on is simple: reject, correct, redraft, resume, every time, without exception for how many contacts the campaign has or how long it's been running. If a campaign ever stops after a rejection going forward, treat that as a genuinely new issue worth reporting rather than an expected part of the workflow.
Availability
This applies to any campaign with a rejected draft, once the underlying instructions have been updated.