TL;DR We’ve been testing Meta’s new generative AI ad tools over the last few weeks. It’s ambitious, but not production-ready yet. Some features save real time. Others miss the mark completely.

If you’re managing spend right now, here’s what you need to know.

The Insider Take

Some advertisers have gotten real access to Meta’s new AI ad generator. We’ve been testing it on a select number of low-spend accounts.

Early verdict: it’s interesting tech, but the creative outputs are still hit or miss.

The good: Background replacements, product crops, and layout variations that keep creative fresh without touching the core brand message.

The bad: Generative edits that make no sense. We had one mens apparel ad turned into a lifestyle photo of elderly women in sweaters. That version actually made it into the auction.

So yes, AI can speed up creative production, but right now it still needs human supervision.

What’s Working

  1. Flex-format ads This is the real MVP right now. You can load multiple variations of creative (images or videos), and Meta will automatically optimize which ones get shown. It’s simple, scalable, and performs consistently.

  2. Catalog-based edits Meta’s AI does a solid job at automatically cropping, swapping backgrounds, and resizing for placements. It saves time and keeps visuals sharp across devices.

  3. Small automations Auto-resizing, headline testing, and layout adjustments are all fair game. These help speed up testing and reduce manual creative setup.

What’s Not Working 🚫

  1. Full generative AI ads Still too unpredictable. The AI often reimagines brand tone or visuals incorrectly, and it’s not reliable enough to use in live campaigns. If you want to test, do it with a small budget in a low-risk environment.

  2. Late rollout timing Launching this right before Q4 made it tough. Most advertisers already have campaign structures and offers locked, so introducing new features now adds unnecessary risk. Save your major AI testing for Q1, when performance volatility matters less.

  3. Creative mismatches Duplicated ads sometimes carry over old copy or outdated promo text. Always check your dynamic placements and ensure creative updates have synced properly before launch.

The Bigger Picture

Meta is trying to automate creative generation at scale, and that’s the right long-term move. But the system is still catching up to the level of nuance that real brand creative requires.

For now, think of AI as a creative assistant, not a creative director.

Automation can make your workflow faster, but it can’t replace instinct, emotion, or brand storytelling. Those things still need people.

Action Plan for Brands

  • Skip full AI-generated ads for now. Keep testing offline or in small budgets

  • Use flex-format for creative testing and scaling

  • Take advantage of Meta’s catalog and background tools to speed up production

  • Audit all ad duplicates before launch

  • Revisit AI creative testing in Q1 when stability returns

Meta’s AI roadmap is heading in the right direction, but it’s early. The smart move is to experiment without overcommitting.

Because when this tech finally clicks, it’ll change how we make ads, again.

Keep Reading