FACTUAL ACTIVITY RECORD · An activity record based on work that took place
Building the AI Talent Editorial publishing workflow
Connecting source intake, drafting, review, approval, and publication through one state model
Purpose
AI-generated articles needed an end-to-end path to publication with human review. The design had to separate draft generation from publishing authority and expose only published articles to readers.
Implementation
Data structures and operations were added for sources, runs, articles, revisions, and reviews. Automated work is limited to draft submission, while the human admin interface handles review, approval, publication, and return for changes. Approved or published work can return to review when editing resumes.
The admin interface gained editorial status, source management, and article editing views. Public surfaces gained an article index, individual articles, and a feed. Markdown passes through a safe renderer, and every public path shares the published-only rule.
The primary diff confirms the retrieval boundary that filters by publication state. Omitted lines assemble relations; the shown condition matches the cutoff implementation.
Rendering diagram…
What was confirmed
Verification results
Permissions and tests confirmed that automation has no publication action and stops at draft submission.
State transitions, published-only retrieval, safe Markdown rendering, and feed generation passed the workflow tests and build.
At the cutoff, the admin-to-public workflow was applied to the target environment and could serve published articles.
Basis for completion
Completion required an implemented and verified state model from source through publication, with the human authority boundary and public delivery path intact.