Ad creation is getting faster, no question. After watching Marketing Against the Grain put Replit 4’s new ad tool through its paces, I’m convinced the hype misses a hard truth. AI will get you moving, but it won’t save you from the grind.
As a marketer and builder, I want quick wins as much as the next founder. But I also care about ads that actually convert. My take: AI ad builders are great for draft copy and search formats, yet weak on visual creative and brand fidelity. That gap matters if you have a real budget or a real brand.
What Kipp Bodnar Got Right
Kipp Bodnar’s test drive hit a core point. Good prompts up front save cash and time. He showed how crafting the brief in Claude, then feeding it into Replit, beats brute-force iteration inside the design tool.
This is one of the worst ads I’ve ever seen. Please redo it.
That blunt line wasn’t performance art. It was a diagnosis. The copy sometimes worked. The visuals often didn’t. Logos were wrong. Color choices drifted. Some ads looked like they were built by a bot that never met the brand.
Set up in minutes, not months… Sign me up. That’s good copy.
Here’s the split: text-heavy placements shine, especially Google search ads. Visual placements? Not there yet without strong direction and manual cleanup.
The Real Friction Isn’t Speed—It’s Standards
It’s easy to launch an app now. The hard part is finding fit. Ads are the fastest learning loop we have, so quality matters. Kipp pointed out that generative tools help you go from zero to one. I agree. But zero to one isn’t the goal—profit is.
When the tool produced a LinkedIn ad with the wrong HubSpot logo and random color choices, that wasn’t a small miss. That’s a broken signal to your market. Visuals carry trust. If the brand mark is off, you’re burning dollars.
Search was another story. The Google ad mockups were clean, clear, and helpful for planning tests. Headlines like “Set up in minutes, not months” and “65% auto-resolved” hit intent and promise. That’s where AI already feels strong: short copy, structured formats, and quick variations with scores for A/B tests.
My Playbook After Watching The Demo
I build and market products for a living. If I were using these tools tomorrow, here’s how I’d run the stack:
- Craft the brief in Claude first; add proof points, audience, and channel rules.
- Feed Replit a strict do-not-do list: banned styles, off-brand colors, logo files.
- Start with Google search. Prove the hook before you fight imagery.
- For LinkedIn and Instagram, supply approved images and brand assets up front.
- Expect multiple revs. Budget $20–$40 in credits for a solid first batch.
This isn’t “set and forget.” It’s “set, test, and tune.” Kieran Flanagan and Kipp talk about running loops—express, tailor, amplify, evolve. That’s the right mindset here. Once you have early data, the next rounds get faster and sharper.
Counterpoint And Why It Falls Short
Some will say AI tools make creative teams optional. I don’t buy it. Great creative is still human-led. The models don’t know your taste, your risk profile, or the small brand details your buyers judge in a blink. They can draft. They can’t decide.
Another argument is that design tools like Canva can fix the misses. True. But that confirms the point: you still need eyes, assets, and standards to cross the finish line.
Where This Leaves Marketers And Makers
I like Replit’s direction. The Canvas view is helpful. The speed is real. But I’m not giving my credit card to an ad tool and walking away. AI is the accelerator, not the driver.
If you’re early or under-resourced, this workflow will level you up. You’ll get working copy, mockups, and a plan to test. If you’re brand-heavy or performance-obsessed, expect to guide the system hard and fix creative by hand.
Here’s my bottom line: use AI to learn fast, then use taste to win.
Do This Next
Want better ads without lighting money on fire?
- Write a tight brief in Claude with real numbers and brand rules.
- Test search ads first. Lock the hook. Then scale visuals.
- Add a clear not-to-do list in every ad request.
- Iterate in short loops and keep score. Kill weak angles fast.
I’m Joel Comm. I’ve built and marketed online since dial-up. Tools change. The job doesn’t. Ship, test, learn, and raise your standards. The market will reward the ones who do.
