Three responses, three different meanings
- Act on it or complete it — the prepared work is right
- Edit the draft first — close, and correcting beats explaining
- Ignore with a reason — wrong in a way that will happen again
Understand what happens when you approve, edit or reject — and why your rejections are the most useful thing you produce in week one.
The next lesson asks you to work your own queue for a week. This one is what makes that exercise worth anything, because the difference between a useful week and a wasted one is whether you respond to items deliberately or clear them.
The three responses mean genuinely different things. Approve says the prepared work is right. Edit says it is close and the correction is quicker than an explanation. Reject says it is wrong in a way that is likely to recur, and that is the one that carries information back.
The three meanings are the useful way to think, but they are not three buttons. Each card carries one primary control that is the decision itself — write the message, enrol the account, mark the work complete — and an overflow menu holding the ways out. A message card opens a blank panel: you write the message, send it from your own email, and the panel saves it to the action. RevTech does not send it. Editing a drafted message happens in AI Messaging, where a sequence draft is edited and approved before you copy it out and send it yourself. Rejecting is Ignore, and it asks you for a reason. Handing it to someone else is Escalate.
One class of item behaves differently and it is worth recognising on sight. Sign-off on a drafted campaign, a new version of a play, a new analyst question or a drafted artifact arrives as a review card with three recorded answers: Approve, Request changes and Reject. Request changes means "not like this" and Reject means "this should not happen", and both ask for a reason that is stored with the decision, so the author knows what to do next. You only see these answers if your role holds the permission for that kind of sign-off. Without it, read the card and raise it with whoever owns the motion.
The habit worth forming now, while everything is new and you are noticing things a month of familiarity will make invisible, is rejecting with a stated reason. A rejection with a reason is a correction the system and the people configuring it can act on. A rejection without one says only that something was wrong somewhere, which nobody can do anything with.
For work routed to review, approval is a real gate: the action stays pending until it is approved. Read the evidence, edit or reject where needed, and understand which other actions your configured permissions allow to run. Review rules must be explicit for each workflow.

Now watch the video
This lesson is the orientation. The material itself lives on its own page, kept in one place so it stays current everywhere it is referenced.
You should be able to answer each of these from memory before opening it. Recalling the answer is what makes it stick; recognizing it when you read it does not.
Approve: the prepared work is right. Edit: it is close and correcting is faster than explaining. Reject: it is wrong in a way likely to recur, which is the response that carries information back.
It says only that something was wrong somewhere, which nobody can act on. A rejection with a reason is a correction the people configuring the agents can actually apply.
Approval is a real gate, not a confirmation dialog. That is what makes the queue worth reading rather than clearing.
In your first week you are not just doing your job — you are calibrating the system. A rejection without a reason cannot do that.
Every item carries why it was surfaced. Approving without reading it is how a queue becomes a formality.
Explore how agents prepare work across your revenue teams, and how people review it, guide it and decide what happens next. Fictional data. Real product.
Explore the demo