GrowthPilot checks every draft before a person reviews it, and keeps a record of what the draft was written from. Neither replaces human review; they make it faster and better informed.
Provenance
Each revision records the Brand Profile, the brief and the stored evidence items it was written from, and whether it was written from measured evidence or from a manual brief only. Open a draft to see its provenance. Evidence references keep their source and the date they were collected.
The six quality gates
| Gate | Checks | Required |
|---|---|---|
| Evidence and context | The draft has stored evidence and a complete brief (objective, audience, keyword, call to action). | No |
| Structure | Headings, length against the brief, internal links and a call to action. | Yes |
| Brand and voice | The Brand Profile was used and prohibited language is absent. | Yes |
| Facts and sources | Measured claims match stored evidence; claims needing verification, unsupported superlatives and statistics without a declared source are flagged. | Yes |
| SEO and schema | Title and description lengths, keyword use without stuffing, and structured data where it applies. | No |
| Claims, legal and policy | Regulated or risky wording, such as guarantees, medical or financial claims, is flagged for review. | Yes |
Each gate is pass, needs review, fail or unavailable, with the reason. A failed required gate blocks Ready to publish until the draft is fixed. Advisory gates inform the reviewer but do not block.
How to run the gates
- Open the draft and run the quality gates from the draft, or
- ask the Agent to Write, review and fact-check content for the client. The Agent runs the gates and the fact-check and leaves the draft for your review.
The report is stored with the revision, so reviewers see the same result.