After watching Marketing Against the Grain break down AI-made ads, I came away with a clear view: AI can speed up ad creation, but it cannot replace taste, brand sense, or channel savvy. As someone who has built online businesses and campaigns for decades, I see the promise. I also see the traps.
Kipp Bodnar and Kieran Flanagan stress-test an AI ad workflow using Claude to craft a prompt, then Replit to generate creative. The results swing from useful to ugly. That swing is the story. AI is a force multiplier only when a sharp human edits, directs, and sets guardrails.
The Core Argument: Treat AI Like A Junior, Not A CMO
The strongest moments came when they called out obvious misses. The kind a machine won’t catch without help. That starts with the blunt opening line:
This is one of the worst ads I’ve ever seen.
They run a LinkedIn concept and like the lead data point—
77% fewer tickets, zero new hires.
—but the visual falls apart. Brand mistakes follow:
First of all, this isn’t the HubSpot logo.
And the verdict is honest:
This ad is really bad. Please redo it.
Then, a flip. A Google ad works because it asks less of visuals and more of copy. They also pull a clean message for Instagram:
You’re not a chatbot, your AI support teammate.
Their punchline lands:
These tools are really great… If you’re a really good designer, they might be slower than actually just ripping and doing some of this stuff in Canva.
I agree. AI shines for drafts, options, and speed. It stumbles on brand fidelity and layout laws. If you hand over your budget to a bot, you pay twice: once in wasted spend, again in lost trust.
What Worked—and What Failed—With AI Ads
I saw three clear patterns in their test that match my own campaigns.
- Channel fit matters. Text-first formats (like search) are safer. Design-heavy formats expose AI’s taste gap.
- Facts beat fluff. The “77% fewer tickets” line hooks. Numbers focus the brain.
- Brand accuracy is non‑negotiable. Wrong logos and layout sins will tank performance and hurt reputation.
That mix tells me AI is good at giving you a starting point and bad at keeping your brand intact. You need human checks before money hits media.
My Playbook For Using AI Ad Makers
I’ve built and sold with every tool under the sun. Here’s how I’d run this stack without burning cash.
- Write the brief first. Define the audience, pain, proof, and one action. No tool will fix a fuzzy goal.
- Engineer prompts like a boss. Use Claude to build role, tone, guardrails, and must‑include claims. Ask for three distinct angles.
- Lock your brand assets. Upload the correct logo, hex colors, and fonts. Force the AI to use them.
- Match the channel. For search, push copy. For LinkedIn and Instagram, keep text off images and respect safe areas.
- Audit for legal and truth. Claims need sources. If you can’t prove it, cut it.
- Refine like a pro. Pick the best line, rewrite it tighter, and simplify the layout. Less ink, more impact.
- Test small, scale fast. Launch micro-budgets across two to three variants. Kill losers in 48 hours. Feed the winner fresh creatives weekly.
This workflow treats AI as a quick idea machine, not a decision maker. It saves time without risking your brand.
Where I Push Back
Some will say AI will soon out-design us. I don’t buy it. Taste is learned in the trenches—through losses, feedback, and pattern memory. Tools don’t live with the cost of a bad quarter. People do. Until machines can honor brand, context, and ethics on their own, you need a human at the helm.
That said, ignore these tools and you’ll move too slow. The edge goes to teams who draft with AI, edit with care, and ship more tests than their rivals.
Final Take—and Your Next Move
AI ad makers are worth using—but only under strong human control. Borrow speed, not judgment. Build briefs, demand proof, and protect your brand like it’s day one.
Start this week. Pick one campaign. Use AI for three concepts. Edit hard. Launch tiny tests. Keep the winner. Repeat. Let AI push volume, while you keep the taste.
