The AI Ad Creative Playbook

A complete, step-by-step system for producing hundreds of ad creatives per day using AI tools. Based on analysis of 105,261 tweets across 244 accounts, with specific pipelines extracted from operators generating $2K to $100K+ per day in the ecommerce ad creative space.

This is the system. Not theory. Not "what could work." This is what operators like @spwfeijen ($2K/day selling AI videos to brands), @FynCas (claims $430K/month with AI-generated ads), and @alexgoughcooper (creative partner for 100+ DTC brands through Adcrate) are actually using.

---

Chapter 1: The Full Stack

There are six core tools. You need all six. Total monthly cost: under $100.

Layer 1: Avatar and Face Generation

Nano Banana Pro ($13/month)

This is the foundation. Nano Banana generates AI avatars with motion control, allowing you to create realistic "UGC creators" that don't exist. You own every avatar you create. No model releases. No talent fees. No scheduling conflicts.

Nano Banana appears 187 times in AI UGC content tweets alone, with 363 total mentions across all categories. It is the most-cited tool in the AI UGC production pipeline by a wide margin.

What it does: generates a photorealistic human face and body that can be animated, lip-synced, and placed in any environment. You control the age, ethnicity, gender, expression, clothing, and motion.

Key settings for ad creative:

• Resolution: Maximum (1080p minimum for ads)
• Motion type: Natural (avoid the "uncanny valley" stiffness)
• Expression range: Moderate (too expressive reads as fake)
• Lighting match: Set to match your intended scene lighting

Layer 2: Video Generation

Kling 2.5-3.0 ($25/month) or Veo 3 (Google AI subscription)

These are your video engines. Kling dominates with 768 mentions in the AI UGC category alone and 901 total mentions. Veo 3 is rising fast with 350 total mentions and Google's infrastructure behind it.

Kling excels at: lip sync accuracy, motion control, cinematic lighting Veo 3 excels at: longer clips, more consistent character appearance, better environmental generation

When to use which:

• Kling: talking head UGC, testimonials, product demos (anything requiring precise lip sync)
• Veo 3: lifestyle scenes, environmental shots, product-in-use sequences (anything where the setting matters more than the face)

A full anime-style music video costs approximately $42 total to produce using Kling ($25) + Nano Banana ($13) + Midjourney (~$2) + SUNO (~$2). This cost structure was verified from a Japanese creator's detailed breakdown shared via @Kling_ai.

Layer 3: Voice Generation

ElevenLabs ($22/month for Creator plan)

ElevenLabs handles voice cloning and AI voiceover. You need this for every talking head ad, testimonial, and narrated product demo.

Critical technique: Create 5-10 distinct voice profiles. Each voice becomes a "creator" in your roster. Match voices to avatars and keep the pairing consistent. This builds familiarity if you're running ads from the same "creator" across multiple campaigns.

Settings that matter:

• Stability: 65-75% (too high sounds robotic, too low sounds drunk)
• Similarity boost: 80-90% (keeps the voice recognizable)
• Style: 30-40% (adds natural variation without going off-script)

Layer 4: Script Generation

Claude (Sonnet or Opus)

Claude is the brain. With 2,630 total mentions across all categories (more than any other tool in the dataset), Claude is the most-used AI tool in the ecommerce operator space.

@alexgoughcooper's specific workflow: "Give Claude the reference ad and ask for a brief to recreate it in extreme detail. Whatever it gives you, respond: this is a 4/10 in terms of graphic design detail. I need a 10/10. GO INTO EXTREME DETAIL. Then paste that into your image generator."

Claude writes your ad scripts, analyzes competitor creative, generates prompts for your image and video tools, and handles the strategic layer of your content operation.

Layer 5: Assembly

MakeUGC v2 (subscription varies) + CapCut (free/pro)

MakeUGC is the template layer. It takes your avatar, voice, and script and assembles them into complete UGC-style ads. Both @spwfeijen and @FynCas name MakeUGC as the assembly tool in their pipelines.

CapCut handles final editing: trimming, transitions, text overlays, captions, and export settings.

@spwfeijen's claim: "Clawdbot + MakeUGC V.2 = 550 videos per day. Fully-realistic UGC ads with cinematic lighting, human motion, perfect pacing."

Layer 6: Static Image Generation

Midjourney (~$10/month) or ChatGPT (image generation via Plus subscription)

For static ad creatives (single images for feed ads, stories, catalog listings). @alexgoughcooper confirmed: "Claude for prompting + ChatGPT for static generation is insane."

Midjourney produces more photorealistic product photography. ChatGPT is faster for iteration and better at following specific layout instructions.

---

Ch