Ad Variant Production.
One approved master concept becomes 10-50 variants across every placement format — 1:1, 4:5, 9:16, 16:9 — the same day. The AI employee runs 3-5 fresh variants per week per active campaign to beat creative fatigue, and every asset passes brand QA and a named human approver before it ships.
How the AI employee does it
(from the Designer playbook)Multiply the master — one approved concept rendered into 10-50 sized variants across all placement formats
Self-QA every asset — brand-token compliance, spell-check of rendered text, and spec validation before humans see it
Run the approval loop — proof sheet with per-asset approve/reject/comment; rejects revised, target ≤2 rounds
Push approved creative — finished ads uploaded to Meta as paused ads for the client to launch
Tools on the job
(official APIs)fal.aiBannerbearMeta Marketing API
Full stack on the Designer knowledge base.
Measured by
(the number that matters)Turnaround per request — same-day versus the 1-3 business-day human benchmark, sustaining 3-5 new variants/week
Every action logged and reviewable — how trust works here.
Related tasks
(the staff work together)Put this on someone's desk
(alpha)Kenji (Ext. 106) handles ad variant production as part of the Designer seat — flat from $499/agent/mo, one agent per dedicated machine. Build the staffer.