Every creative format species we could build an ad in — 57 of them, in 9 families, curated from an 82-species practitioner taxonomy plus competitor evidence with verified run dates. This is an inspiration library, not a shortlist: nothing here is picked, and the off-register entries stay on the wall so you can see we considered and rejected them — not that we missed them. Rows marked ● carry a real ad from the wild — actual creative, dated. The five formats already produced for hook #19 are marked.
The creative pretends to be a document, a message, a calendar — something she'd see anyway. This is Halbert's A-pile in visual form: it earns the read before it's recognized as an ad, and native-format tests show ~35% higher view completion than non-native. The line that governs the whole family: mimicry of a document is honest, mimicry of a system alert is not (Perplexity's fake warning-screen ads are the cautionary tale — borrowed trust is spent trust).
No artifact pretense, no image: set type says one thing with total confidence. The most durable proof in both categories lives here — Replit's plainest text ad has run 408 days, and Teachable's plainest has run 1,740 days and is still live. Plain beats clever is this family's law, and it's the most premium-compatible family on the wall: restraint reads as confidence.
Proof by demonstration is the one move info-marketers cannot copy — they have nothing to show. It's also where the congruence mechanic is native: ad promise = prompt output, one artifact. And today's sweep confirmed something bigger: no coach-economy advertiser runs a single demonstration or prompt-box ad. Replit and Lovable built their programs on this family; in our category it is completely unclaimed. The constraint is real though — most species here need product capture, and the best moment (mid-build) sits behind the onboarding scrim.
Three sourcing tiers, already ruled: us first (days — the product does the talking), paid coach-creators (weeks), real customers (months, permission + likeness release + payment, now cleared). The register law for the whole family: unpolished beats produced on Meta — audience ad-skepticism is the enemy and polish triggers it. In-category proof: Teachable kept one UGC creator video in rotation for 588 days.
Proof is mandatory on cold traffic — that's doctrine, not preference — and customer names in creative are now cleared with permission and payment. This family makes the evidence itself the ad. The register line: documentary, never altered, never stacked into hype. Verified roster only.
Everything here is desk-producible with the existing Remotion pipeline, governed by the rate law — each species has a different governing rate, budgets in characters or words per second, 38px legibility floor. Note: Remotion needs a paid company license before anything ships as paid media.
The feed's own dialect: content that refuses to look like an ad at all, plus the photography-led species that trade on warmth instead of argument. Cursor proved the lo-fi end works in dev tools. For Circle the line is precise — plain is our register, ugly is not — and several entries are on the wall as a decided no rather than an absence.
The expensive end: real filmmaking where the product enters as the resolution, not the subject. Both flagship proofs in the category are brand-made, not customer-made. In a showcase where quality is the binding constraint, not cost, this family belongs on the wall — it's what "nail the creative" looks like at the ceiling.
Formats where what the ad does — poses a question, answers an objection, sequences an argument — matters more than what it looks like. Small family, high congruence with the funnel: the qualification and intent-protection doctrines both have native ad formats here.