Every AI-written email now passes through a stricter set of checks before it is allowed to send. Claims need evidence behind them, the copy itself is checked for quality, meeting logic has to make sense, cadence rules are enforced, and repetition across a sequence gets caught. The billing receipt for a draft is now tied to that exact draft, not an approximation.
What changed
Before a draft is allowed to send, it now goes through several distinct checks, each aimed at a different way a draft could go wrong. First, any specific claim in the email needs to be backed by evidence, so the AI cannot state something about a prospect or their company that it cannot support with real research. Second, the copy itself is checked for quality, catching drafts that are poorly written, awkwardly structured, or off tone before they reach a prospect's inbox. Third, meeting logic is validated, meaning if a draft references scheduling or a meeting, that logic has to actually be consistent with the campaign's setup, rather than referencing availability or steps that do not match how the campaign is actually configured.
Fourth, cadence is checked, so a draft that would break your intended sending rhythm, whether by going out too soon after a previous touch or skipping a step in the sequence, gets caught before it disrupts the pattern you set up. Fifth, the system checks for repetition, so a sequence does not send the same phrasing or structure to the same prospect more than once, which would otherwise make a multi-touch sequence feel mechanical and obviously automated.
Alongside these checks, the billing receipt tied to a draft is now exact. Previously, the connection between what you were charged and the specific draft that charge related to could be looser than it should have been, making it harder to reconcile spend against actual output. Now, the receipt maps directly to the exact draft it belongs to, so if you are reviewing spend, you can trace it back precisely, email by email.
These five checks run in sequence, not as a single pass and release decision. A draft has to clear each one on its own terms, which means a strong draft can still be held back over a narrow, specific issue, like a repeated phrase from a previous touch, without that flaw needing to be a sign of a broader problem with the campaign. Treating each check separately also makes it easier to tell what actually went wrong when a draft does get held, instead of getting a single generic rejection with no explanation behind it.
How to use it
You do not need to configure anything for these checks to run; they apply automatically before a draft is eligible to send. What you will notice is that a draft failing one of these checks will not go out, and instead surfaces for review or gets held for correction, rather than reaching a prospect in a state you would not have approved.
When you are reviewing a campaign's drafts, especially early in a new campaign, pay attention to any draft that gets flagged. The specific check it failed (evidence, copy quality, meeting logic, cadence, or repetition) tells you exactly what to look at, whether that is tightening your instructions or checking that your research inputs are strong enough to support the claims you want made in outreach for that segment.
When you review billing, use the receipt tied to a specific draft to confirm exactly what you were charged for that email, rather than reconciling against an aggregate number that does not tell you which draft it belongs to.
If you are launching a new campaign for a segment you have not written to before, expect to see more drafts flagged in the first few days than you will later on. That is normal: the checks are calibrating against a new combination of instructions, research, and sequence structure, and a slightly higher hold rate early on is the system doing its job, not a sign that something is broken.
Why it matters
A single email with an unsupported claim, broken meeting logic, or repeated phrasing across a sequence can cost you a prospect's trust, and there is no way to get that first impression back once it is sent. Catching these problems before send, rather than after a prospect notices them, protects the reputation of every campaign you run and every future email that prospect might otherwise have been open to reading.
Tying the billing receipt to the exact draft matters for a more practical reason: it means you can actually audit what you are paying for, email by email, instead of trusting a summary number that hides the detail behind it.
Together, these checks and the tighter billing receipt are meant to make send-time the last line of defense rather than the first, so the bulk of quality control happens before a prospect ever sees the result, and what you review as a customer of the system is spend you can trace back to specific, accountable output. That is a different relationship with the product than trusting a black box to get it right on the first try.
It also means the burden of catching problems shifts away from you reading every draft by hand before it sends. You still can, and for a new campaign it is worth doing, but the checks now carry the routine, repetitive part of that responsibility, freeing your review time for judgment calls the checks are not built to make.
Availability
These validation checks and the exact draft-level billing receipt apply to AI-written emails across your campaigns now, covering first-touch messages, follow-ups, and replies alike, with no configuration needed on your part to turn any of it on, whether the campaign has been running for months or launches for the first time tomorrow, and regardless of how many contacts it is currently drafting for, whether that is a handful of contacts or several thousand spread across many active sequences.