With adomate you create ads from data, at scale. Take data from your Meta ad account, the ad library and reviews from Trustpilot and Amazon. Build creative research workflows that create branded ad concepts, the way you want. Fully traceable to their original data trigger, no blackbox.
Free Options
Launch Team / Built With
Render: Get up to $100K in startup creditsCloud infrastructure for founders who want to move fast and NOT break things
Promoted
Hey Product Hunt 👋
Simon here, co-founder and CEO of Adomate.
Before Adomate was a self-serve platform, Lucas and I ran it as a service, building and delivering Meta ad creatives by hand for DTC brands across Belgium and the Netherlands.
We wanted to feel the problem first:
Strategists losing 60-70% of their week to research instead of making
Rebuilding context from zero every Monday
No system connecting last week's winner to what gets briefed next
Creative is the last lever you actually control on Meta now that targeting and bidding are automated, and most teams still run it like it's 2019.
The solution
We built Adomate. It puts your performance data, competitor intel, and consumer signals into one workflow you build and own. A few things it's already done for early teams:
→ Increased number of winners and ROAS
→ Volume: helped Nomige to meet creative demand and deliver the number or creatives required for testing
→ Golden nuggets: automatically found usable insights in thousands of customer reviews
How it works
1. Feed in your performance data, brand brain, competitor ads, reviews, and consumer signals. One place, not ten tabs, so research time drops to minutes.
2. Build the workflow once. Every data point after that becomes an ad concept, so you stop starting from scratch every session.
3. Like the concept, skip it, or refine it in plain language. Point out a reference if you've got one. Every choice trains Adomate and saves a version, so getting from 80% to 100% takes minutes, not a new brief.
Why it's different
No black box. Every concept you see traces back to the exact data point that triggered it, the competitor ad, the review line, the performance metric. You can always see why.
Adomate doesn't replace your judgment, it amplifies it.
Who it's for
Creative strategists and performance marketers at DTC and e-commerce brands running Meta ads, plus agencies managing creative across multiple accounts.
If briefing your designer feels slower than it should, this is for you.
Special for the PH community
→ 30% off your first 3 months: PHMONTHLY30. 300 codes.
→ 50% off your first year: PHANNUAL50. 100 codes.
Apply the code at checkout. Valid until Friday 11:59pm PT, or until the codes run out.
Our ask
If you're running paid social, tell us what's broken in your workflow right now. That's the input that shapes what we build next.
👉 Get started for free to see what you can do with Adomate.
@lucas_desard and I will be here all day. 👇 Big thanks to @rohanrecommends for hunting us 🙏
@lucas_desard @s_logghe Good luck with the launch!
@lucas_desard @sven_de_meyere1 Thanks Sven!
@lucas_desard @s_logghe Trying to answer your q. i think the biggest nightmare for us is turning winning static ads into video briefs. Huge congrats🙌 on the launch supported.
@lucas_desard @rohanrecommends @priya_kushwaha1 Hi Prya, thank you for your input. On our roadmap: Video briefing generation from all types of sources: statics, videos, reviews, etc. Stay tuned!
@lucas_desard That's great 🙌
Best of luck with the launch buddy! @s_logghe
@yakha88 Thanks Yannick!
@s_logghe the traceability is the part that gets me .. every concept pointing back to the exact review line or metric that triggered it 🙌🏽 that's the difference between a tool a strategist trusts and one they quietly stop opening
@lilhadi Hi Haider, exactly. After months of working alongside performance marketers, we learned that handing control over to an LLM is not what they want. Marketers are trained to run experiments, draw conclusions, and build knowledge incrementally. We give them the insights and control to do just that, by working with ground truth data, letting them set up their own workflows, and building trust and understanding from the ground up.
@s_logghe great approach, gonna give it try 💯
@lucas_desard @rohanrecommends @s_logghe From the buyer side, "at scale" is where these usually break for me. Generating 40 ad variants is easy; knowing which three to fund is the call agencies charge for. Does it close the loop with spend data and kill the losers, or just produce creative?
@lucas_desard @rohanrecommends @artem_fedorovich Hi Artem, good question. Spend data is one of the 3 data sources we work with (next to ads from other brands and consumer reviews). So performance data is directly linked to the creative process and variants only get made from proven winners.
The traceability is the strongest part here. AI can generate endless ad ideas, but without knowing whether a concept came from a winning campaign, a competitor pattern, or an actual customer review, it is hard to trust or learn from the output. As someone doing more launch and marketing work, I can relate to rebuilding context across too many tabs and starting the creative process from zero every time. Curious how quickly Adomate learns a team's taste from likes, skips, and refinements before the concepts start feeling genuinely tailored :)
@andrasczeizel Hi Andras, thank you for your comment. As always with data, the more the better :). So the more feedback Adomate can capture, the faster it goes. Users can also directly edit brand and taste settings themselves at the source.
Congrats on the launch. The service-first route makes sense here. You usually learn much more by doing the work manually before trying to automate it. I'd like to understand is how Adomate decides what should influence the next brief. Performance data can tell you what won, but not always why it won. Can the team challenge the system’s reasoning or add context when a creative worked because of timing, an offer, or something happening in the market? That seems important if the goal is to avoid turning last week’s winner into next week’s template.
@os_ishmael Hi Os, great question. As a data-first company, we're extremely careful about attributing the success of an ad to a specific cause. It would be easy to just ask an LLM and get a convincing-sounding answer, but in reality, you'd be attempting to reverse-engineer one of the most sophisticated commercial algorithms in the world. We'd rather show users the facts and all available data, and let them apply their own experience and market knowledge to draw the right conclusions. Adomate is built to extend the marketer, not replace them.
@ridhwikvinod Hi Ridhwik, thanks for your question. Other platforms and data sources are definitely on our roadmap. DTC was our obvious first target, but we can and will replicate our methods to B2B as well.
Congrats on the #1 launch 🎉 The part that stands out to me isn't "AI generates ads" — it's the "fully traceable to the original data trigger, no blackbox" framing. Most gen-creative tools hand you an output with zero lineage, so you can't tell why a concept was made or how to steer the next one. Curious how granular that trace actually is: when a concept links back to a data trigger, does it point to the exact source item — this Trustpilot review, this ad-library creative — and can I act on it, e.g. "this hook came from these 3 reviews, regenerate weighting them differently"? Basically: is the traceability a provenance view, or an actual control surface you steer generation with? Really nice work either way 🙏
@akbar_b Hi Akbar, great question. At the moment, generation starts from a single data point, in your case, a single review. The system identifies the one review that best matches your self-defined criteria and builds from there. Steering from higher-level, aggregated insights is on our roadmap, but for now we work with one data entry at a time.
@s_logghe got it, thanks. Seeding from one review does make the provenance really clean — you can always point at the exact source behind an ad. The thing I'm curious about: one review is one voice, so the best-matching one could be an enthusiastic outlier rather than a typical customer. When you add the aggregated-insight steering later, do you think you'll keep that clear line back to a specific source, or does pulling from the aggregate make it fuzzier which input actually drove a claim?
@akbar_b It will definitely make it more fuzzy, and don't forget we're talking advertising here, so you might want to search for those outliers (attention grabbers) rather than optimizing for the mean.
@s_logghe Good reframe — I was importing an 'outlier = noise' instinct from ML, but for ad copy the outlier voice is often the point. Wonder if 'best match' should split into representative-of-the-mean vs most-distinctive-voice modes, or if that's already implicit today?
Simon, you asked what is broken, so here is mine. I run paid acquisition in healthcare, and the thing that breaks is upstream of creative.
The loop learns from whatever the ad account calls a winner, and in my market that label is wrong. The conversion I can actually fire is a trial signup, and trials and paying customers are not the same population, so my real cost per paying customer came out roughly double what the platform reported. A creative engine trained on that will confidently scale the concept that produces the cheapest signups, which is not the one that produces revenue.
Can Adomate learn from an outcome uploaded back later, a paid conversion at day 30 for instance, or does winner mean whatever Meta says it means?
@clemente_lopez1 Great question, Clemente. This is actually something Meta already supports. You can send your downstream business events (e.g. a paid subscription 30 days later) back through the Conversions API, so Meta can optimize for the outcome that actually matters, not just the trial signup.
If you have enough paid conversion volume, I’d recommend optimizing directly for that event.
If volume is limited, it can make sense to optimize for an earlier, high-quality event, ideally the best early signal for a conversion later (e.g. high product usage in the first few days after sign up) until Meta has enough signal.
@lucas_desard that closes the loop on the platform side, thanks. What I cannot close is volume: paid conversions in my niche land in the low tens per month, so the optimizer would be learning from a trickle and I stay on the proxy event anyway.
And the proxy you describe is the one I have to weaken on purpose. My best early signal is clinical usage in the first days, and shipping that back means telling an ad network something about a clinician's patient work, so what leaves my side is a coarse delayed flag with no detail in it.
Does attaching value to the paid event buy enough signal to offset a low count, or does Meta's model only respond to event volume?
@clemente_lopez1 Great point. You’re describing one of the hardest problems in performance marketing: finding the right event to optimize for. In some industries that’s obvious, but in healthcare SaaS it’s not.
For your clinical usage example, you don’t need to send Meta any confidential information. You can define an internal threshold based on historical data (e.g. users who complete certain actions in their first few days are much more likely to become paying customers) and only send a delayed conversion event once that threshold is met. Meta only sees that the event happened, not why.
As for the low volume, it does make learning slower, but those signals compound over time. Another avenue worth testing is optimizing for a different proxy higher up the funnel that gives you more volume while still correlating well with paid conversions. In the end, performance marketing is all about experimenting, measuring, and doubling down on what proves to work best.
I run paid social on the buying side and support on the receiving side, so the thing that is broken for me sits downstream of yours.
Golden Nuggets is the feature I would watch. The review lines that make the best ads are almost always about an experience rather than a product. Arrived in two days. They replaced it, no questions asked. That is exactly why they sound authentic, and it is also the moment the ad stops describing the product and starts making an operational promise.
The trace tells you where the claim came from. It does not tell you whether you can still keep it. That review was true for one customer, in one country, under last year's courier contract. Run it as an ad and you have made that promise to everyone who sees it.
And it does not fail where you are looking. ROAS goes up, because it is a good promise. It shows up two weeks later in the inbox as "your ad said two days", and nobody connects that ticket back to the creative that caused it.
What I would want, and you are closer to it than anyone because you already hold the trigger: flag which concepts make a claim about time, price, returns or availability, as against a claim about how the thing looks or feels. The first kind needs one person to confirm it is still true before it runs. The second does not. Everything you need to tell them apart is already sitting in the data point you traced back to.
@jernej_jan_kocica Hi Jernej, your feedback is very true. Within the Review dataset in Adomate, you can slice and dice the data the way you want. Based on the use case you described, you could simply add a column that classifies every review in one of the categories you describe (time, price, returns, availability, product, etc). This way you can ignore the reviews that don't make sense.
@s_logghe That works for picking which reviews to mine, and it is the right first move. The bit I would push on is that filtering them out throws away your best material.
The operational lines are the ones that sound most human, which is exactly why they make the strongest ads. So you do not really want them excluded, you want them routed. Same column, different destination. Product claims go straight through, time and returns and availability go to whoever can confirm the promise is still one you can keep.
The harder half is that the answer is not in the review dataset at all. Whether "arrived in two days" is still true depends on this year's courier contract, not on anything a reviewer wrote. So the column tells you which concepts need an answer, but somebody outside the tool has to give it. Which is fine, it just means the check is a person rather than a filter..
@jernej_jan_kocica Hi Jernej, yes makes sense. Validity of the claims to make is definitely very important. And even more if they change over time. The brand brain contains all information about claims to make. For the use case you describe, an approval flow could be introduced to the person going over the operations.
@s_logghe That is the right shape, and the brand brain is probably a better home for it than the review pipeline, because the claim outlives the review that suggested it.
One thing that would stop the approval becoming another queue: only ask when the claim is new. If someone already confirmed two day delivery this quarter, the next concept leaning on it should not stop. Ask once per claim rather than once per ad, or the approval turns into the thing people click through without reading.
Thanks for taking a critical comment seriously on launch day. Not everyone does..