Nobody has a list of the people they have left hanging, because such a list is only visible from the other side. An agent searches what you sent, joins it against what never got answered, and hands you the ranked list — and sends nothing at all.
The thread you never answered is not a memory you can search for. It surfaces weeks later as someone else mentioning it, or a deal going quiet, or a renewal that assumes you agreed to something you never saw.
Your sent folder looks complete. The last word in every thread is yours, which is exactly what a well-kept sent folder means — so nothing in your own view tells you that four of those threads ended with a question you never opened. The information exists, but it lives in the other person's inbox, and there is no view of it you have.
You remember the threads that mattered. What you forget is the mid-sized one from eleven weeks ago — a pricing question, a reschedule request, a document somebody asked for — because remembering it is not currently your job. The threads you have left waiting are, by definition, the ones your attention did not consider worth holding on to.
The failure is always discovered externally: a prospect mentions it in a call, a customer asks whether you ever replied, a deal you thought was progressing goes quiet after a message from you that asked a question and got silence back. That silence is ambiguous — it could mean they answered, or could mean they gave up — and the only way to tell is to go and look, which is exactly the thing that never happens.
Copy it once, paste it into Zaira, and it does the join nobody does: read your sent mail, find the threads that stopped after your message, work out who is waiting and for how long, and rank them. It drafts nothing and sends nothing.
Run once a month, it takes twenty minutes off the list. The useful output is not the sending — it is seeing the shape of your own inbox from the outside, which is usually uncomfortable and always actionable.
You cannot recall these threads, because remembering them was never a task you had. The list comes from joining two sets of data you already have rather than from your judgement about what matters, which is the only reason it contains anything you did not already know.
The prompt forbids sending and drafting outright, not as a default you can override. For a workflow whose entire output is a list of people you have wronged, an agent that cannot write is the only version worth having.
Twenty threads unanswered is a fact about how you work that no amount of intending to reply differently changes. Running it monthly turns it from something you occasionally feel guilty about into a number you can watch move.
The questions people ask about a workflow that reads their sent mail.
No, and the difference matters. A follow-up workflow finds the stalled threads and sends something to them — this one only tells you they exist. That separation is deliberate: sending an acknowledgement three weeks late is often worse than staying silent, and because they are close enough to confuse, the audit page and the sending page should never be run as one prompt. If you want both, run this first and decide what to send yourself.
Because the unanswered thread is not in your inbox. It is a thread where you spoke last, which means it lives in sent. Searching the inbox finds the people waiting on you to read something, which is a different and much easier problem — and one your client already partly handles with a primary tab.
Those remind you about threads you have flagged as follow-ups, so they are limited to the ones you already noticed. This audits all of them, which is the point: the expensive threads are the ones you did not flag precisely because they were not important when you read them, and importance is not the same as whether somebody is sitting waiting.
The prompt drops those, and this is the hardest judgement in the job. A thread that ended with you asking a question and no reply is ambiguous — they may be busy, or they may have decided not to continue — and lumping those in with threads where they asked you something makes the list useless. Expect a fair number of ambiguous cases in the unsure group rather than a false precision on the main list.
Monthly is enough to stop a backlog forming, and often enough that the list stays under about twenty items. More than weekly and you will start working through the list out of guilt rather than judgement, which produces replies you would rather not send. Run it by hand the first two or three times so you learn what is actually on the list before putting it on a schedule.
Only ones where you sent the last message and the thread has not gone cold. Cold outreach threads are the usual false positive, which is why the exclusion list in Step 0 takes senders you never want chased regardless of age. If you have sent something and heard nothing back, that is a legitimate row on this list — often the most useful kind.