Help topic
How drafts are made, checked, edited, and kept safe.
Deloosh combines your brief and campaign goal with the recommended play, audience, confirmed brand context, governed examples, and proof assets. It returns a reviewable email draft, then runs deterministic truthfulness checks before showing you usable copy.
Deloosh blocks performance or effectiveness claims unless a confirmed proof asset states the whole sentence. The refusal appears as unsupported performance claim: “Performance/effectiveness claims require an explicit proof asset claim.”
A refusal means Deloosh could not release a trustworthy draft under the evidence and safety rules that applied to the request. The refusal rationale names the failed check so you can remove the claim, supply the missing proof, or correct the source material.
This message means the selected block cannot safely accept that direct edit. The reliable live routes are to pin an annotation instruction, make a supported direct edit on a compatible block, or regenerate a legacy whole-document artifact into an editable block draft.
Deloosh treats deliverability statements as hard factual claims and blocks them unless a confirmed proof asset states the whole sentence. The refusal appears as unsupported deliverability fact: “Deliverability facts require an explicit proof asset claim.”
Direct design edits use your stored brand tokens and active library images, not arbitrary CSS. The constraint keeps one-off changes from quietly creating a second, inconsistent design system inside an email.
A Draft error means the drafting model did not return a result Deloosh could safely accept. Retry the action; if it fails repeatedly, contact support with the time of the attempt and the error shown in the Workbench.
Deloosh blocks a draft that contains an image with almost no readable body copy, and it labels less-extreme image-forward drafts with an advisory. Add useful text or remove images before trying again.
Deloosh uses real images from your governed brand library; it does not generate product, customer, or brand imagery. A made-up product photograph would be a fabricated claim in pixels, so fresh drafts and edits accept only active images that belong to your workspace.
Deloosh re-runs deterministic safety checks on the block changed by an annotation or direct edit. If the new block introduces an unsupported claim or unsafe markup, the Workbench shows Edit blocked or Patch blocked and leaves the current document in place.
Within the live annotation and direct-edit patch flow, targeted edits are stored as first-class block-document versions and append-only edit deltas. Later targeted changes start from the latest accepted document rather than an unedited original. The stronger promise that a future document-level chat re-draft will automatically preserve every human-edited block is ratified design but is not wired in the current implementation.