Our creative team went from a handful of people to fifteen full-time hires in under a year, and even that growth wasn’t enough to keep up with what Meta’s Andromeda algorithm demands from advertisers right now. That volume problem is exactly why we built Thumbstop, an AI ad creative tool that generates Meta Ads and Applovin images and videos in the time it used to take to write a single creative brief. On a recent episode of the Smart Marketer Podcast, Molly Pittman sat down with Pep Hufen, who runs our agency, to walk through how Thumbstop went from a cold DM on Twitter to a piece of software our own team and outside brands now use to produce hundreds of ads a week.
Key Takeaways
- Thumbstop exists to solve a real internal bottleneck: Meta’s Andromeda algorithm rewards creative volume, and our fifteen-person creative team couldn’t hand-prompt enough variations to keep pace.
- Pep Hufen found his technical co-founder, Vince, by studying who a respected solo AI builder relied on for technical help, then cold-messaging that person’s network directly.
- Customer avatars, not just product photos, are the input that makes Thumbstop’s output usable for direct response, because Meta’s delivery system reacts to environment and context as much as the product itself.
- Brands are already running a 50/50 mix of native and AI-generated creative through Thumbstop, including Smart Marketer’s own pet supplement brand, Longer Together.
What is Thumbstop and why did Smart Marketer build an AI ad creative tool?
Thumbstop is an AI ad creative tool we launched this year that generates images and videos for Meta Ads and Applovin. Pep built it because our own creative team, which grew from a handful of people to fifteen in under a year, still couldn’t produce enough creative volume to satisfy Meta’s Andromeda algorithm.
Before Thumbstop, our creative team was already using AI image generators like Freepik, but they were prompting almost every frame by hand. Pep noticed the gap: plenty of AI tools existed, but very few were actually built for direct response advertising. Andromeda requires so much creative volume that hand-prompting couldn’t scale, so the question became how to use AI in a way that created real leverage instead of just adding another manual step to the workflow.
How long does it take to build and launch an AI ad creative tool?
It took roughly nine months from Pep’s first message to Vince to a full public launch of Thumbstop, with four to five of those months spent getting a production-ready version stable enough to test with outside users. Vibe coding and modern AI development tools make internal prototypes fast to build, but a tool other people can rely on takes considerably longer.
Once the team had a working minimum viable product, they tested it internally first: could Thumbstop generate ads that actually converted for Smart Marketer’s own accounts before anyone else touched it? Only after that internal proof did Thumbstop go out to a small group of mastermind members as beta testers, some of whom reported that early, imperfect assets were already outperforming their existing creative.
Why are customer avatars essential to AI-generated ad creative?
Customer avatars matter because different segments of buyers use the same product for different reasons, and Meta’s ad delivery system can match creative to the right person only if that context is built into the ad. Thumbstop combines brand data, product data, and avatar data so the same product generates a different ad for each distinct customer.
Pep didn’t originally build Thumbstop around avatars. In the first version, users simply wrote one product description covering pain points and benefits. Pep noticed he kept duplicating the same product and rewriting slightly different descriptions for different types of buyers, which told him avatars needed to be a core pillar of the product rather than an afterthought.
Moccasins Canada, a footwear brand and Thumbstop user, is a clear example. Reviews on their best-selling boot revealed three distinct avatars: fashion-focused buyers who care about look and feel, hikers who wear the boots on trails, and hunters who wear them in the woods. Thumbstop now generates avatar-specific ads for each group, and the brand runs thousands of dollars a day on Meta using largely Thumbstop-built creative.
How does Thumbstop build creative diversification for Meta’s Andromeda algorithm?
Since Meta’s Andromeda update, its machine learning models understand ad content deeply enough to match specific creative to specific viewers, which is why Meta pushes advertisers to submit more creative variety rather than relying on manual audience targeting. Thumbstop is built to diversify ads across the four levers that actually influence who Meta shows an ad to:
- Format: producing both images and videos rather than one or the other.
- Ad concept: varying whether the ad shows a before state, an after state, or a full before-and-after transformation.
- Environment: changing the background and setting, such as a forest, a mountain, or a river, which gives Meta a genuinely different signal to match against a different viewer.
- Person and emotion: varying who appears in the ad and how they express emotion, such as a buyer in their twenties versus a buyer in their seventies wearing the same product.
Pep is careful to draw a line here: this is not the AI slop advertisers are rightly wary of, meaning five hundred headline variations or a color swap between otherwise identical creative. Changing the environment or the person in an ad changes who Meta considers the avatar for that ad, which is a meaningfully different thing than shallow visual variation.
How much time and money can an AI ad creative tool save versus manual production?
Using Thumbstop, Pep can generate roughly one hundred meaningful Meta ad creatives in about thirty minutes because the tool generates multiple concepts in parallel. Producing that same volume through a manual creative team, with scripting, shooting, and editing, would take weeks and cost dramatically more per asset.
| Approach | Time to produce 100 ad creatives | Approximate cost | Team required |
| Manual creative production | Multiple weeks | Hundreds to thousands of dollars per creative | Strategists, designers, editors |
| Thumbstop (AI ad creative tool) | 30 to 60 minutes | Roughly $1 per generation on a credit-based plan | One person reviewing and uploading winners |
That cost gap doesn’t mean the manual approach goes away. Smart Marketer will always produce hand-shot, hand-edited creative alongside AI-generated assets, because some ideas are still best executed by a human creative team. The advantage of an AI ad creative tool like Thumbstop is in the volume and iteration speed needed to test and scale, not in replacing craft altogether.
How do you train a creative team to adopt an AI ad creative tool without replacing their jobs?
Training a creative team on an AI ad creative tool works best when it’s framed as freeing up time for strategy, not replacing headcount. Smart Marketer’s creative team already understood direct response principles, like how placing a product in different environments or showing before-and-after states can move performance, they just needed to stop spending fifteen to twenty minutes per prompt trying to get AI output right.
Once the team ran a hands-on workshop with Thumbstop, that unlocked time got redirected toward creative strategy rather than prompting. The tool is also useful for turning already-proven creative into more variations: a team can bring in an existing high-performing static ad and use Thumbstop to animate it into video or generate new concepts that carry the same core idea into fresh formats.
How is Thumbstop being used to launch Smart Marketer’s new pet supplement brand?
Smart Marketer is using Thumbstop to launch Longer Together, a brand-new pet supplement brand built from zero dollars, starting with a dog dental powder called Daily Dental Defense. The launch is running roughly fifty percent native, hand-shot creative and fifty percent Thumbstop-generated creative, including before-and-after and educational ad formats.
The team identified two overlapping customer avatars for the product: owners of aging dogs who want more years with their pet, and owners who already pay for expensive annual dental cleanings and want a lower-cost, preventive alternative. Both avatar types share underlying emotional stakes, which the team is testing across native and AI-generated formats side by side to see how each performs as the brand scales.
What does Thumbstop cost and how does its pricing work?
Thumbstop runs on a credit-based pricing model rather than a flat software subscription, because unlike most SaaS products, every generation carries a real cost since Thumbstop pays an AI model provider for each image or video it produces. The entry tier starts at twenty-five dollars a month, which generates roughly thirty image ads depending on which model is selected.
Anyone testing Thumbstop can sign up at usethumbstop.ai and use the code SMART50 for fifty percent off the first month. Pricing aside, Pep and Molly both frame the tool’s real value the same way: more meaningful shots at goal, at a volume manual production simply can’t match, so a team can spend its saved time on the creative strategy that actually moves performance.
Try Thumbstop and follow the Longer Together launch
Smart Marketer will be tracking Longer Together’s ad performance over the coming months, comparing how the native and Thumbstop-generated creative perform against each other as the brand scales spend on Meta. If you want to see what an AI ad creative tool built specifically for direct response can do for your own ad account, head to usethumbstop.ai and use code SMART50 for 50% off your first month.
