Marketing Against The Grain Can AI Actually Make Good Ads? Replit Ad Maker Review
19 snips
Apr 2, 2026 A live test of Replit’s AI ad maker explores whether cheap, fast tools can create standout ads. It dives into prompt-building with Claude, smarter iteration to save time and credits, and the challenge of making LinkedIn, Google, and Instagram creative actually look good. Copy impresses. Visuals stumble. The real drama is in the revisions.
AI Snips
Chapters
Transcript
Episode notes
Refine Prompts Before You Burn Replit Credits
- Refine the ad prompt in Claude before generating assets in Replit to save time, credits, and rework.
- Kipp uses Claude Opus 4.6 to research HubSpot's customer agent, write a massive platform-specific prompt, then edits it before spending Replit credits.
Edit The Brief Section By Section
- Review each AI-generated ad angle and rewrite the weak parts with your own creative taste before sending the brief into Replit.
- Kipp changes a LinkedIn concept from polished corporate creative to a homemade anti-LinkedIn scroll stopper and has Claude update only that section.
Product Understanding Drives Better AI Ads
- AI ad generation only works well when the model understands the product deeply, not just the channel format.
- Kipp says Claude researched HubSpot's customer agent first so Replit knew the benefits, positioning, and value proposition across LinkedIn, Instagram, and Google.