# How Airpost Works

## The Performance Creative Loop

Every ad we make teaches the next one. Our AI engine and expert strategists work in a continuous loop — analyzing, generating, testing, and learning.

## 1. Analyze Your Assets

Our engine ingests your brand kit, product pages, and past ads. It breaks every asset down shot-by-shot — detecting faces, products, text, pacing, and visual style to build a complete creative fingerprint.

- Brand Kit
- Shot Segmentation
- Visual DNA
- Past Ads
- Face & Product ID
- Text Extraction

## 2. Classify with Taxonomy

Every asset is tagged across 9 creative dimensions. This maps the full creative landscape so the engine knows what's been tried and what's wide open.

- ICP
- Angle
- Tactic
- Format
- Tone
- Pace
- Narration
- Structure
- Funnel Stage

## 3. Generate at Scale

The engine writes dozens of scripts per cycle, scored against a curated Golden Set of top-performing references. Multiple AI models compete — only the best concepts survive.

- 5-50 New ad concepts per week
- Past Ads
- Face & Product ID
- Text Extraction

## 4. Validate with Experts

Human strategists review every concept before it ships. They correct, refine, and explain why — and those corrections retrain the engine. The AI gets sharper with each cycle.

- AI generates. Human validate.
- Corrections become training data.

## 5. Launch & Learn

Ads ship to Meta, Instagram, TikTok, YouTube, and Snapchat. Performance data flows back into the taxonomy — every impression, click, and conversion teaches the engine what works.

- Performance data feeds back into the loop

## 6. Map White Space

The taxonomy reveals which creative combinations haven't been tested yet. Instead of guessing what to try next, the engine identifies gaps and fills them systematically.

- Tested
- White space
- Creative Coverage

## The Compound Effect

### Every Ad Makes the Next One Better

Volume creates data. Data creates speed. Speed compounds learning.

- Analyze
- Classify
- Generate
- Validate
- Launch
- Learn

## Brand Maturity

### Less Human Input Over Time

As the engine learns your brand, human oversight scales down and creative quality scales up.

- **Month 1**: Onboarding — Engine learns your brand. High human review.
- **Month 2-3**: Calibrating — Taxonomy fills in. Corrections decrease.
- **Month 4-6**: Accelerating — Engine suggests concepts autonomously.
- **Month 6+**: Compounding — Minimal oversight. Maximum output.

## See the Loop in Action

A new kind of hybrid creative platform: not AI, not humans - the best of both.

[See a Demo](/content/contact/index.html)