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.
The "fully traceable to their original data trigger, no blackbox" line is the part I'd push on, because I think traceability is necessary and still not sufficient — and it cost us to learn where the gap is.
We build an AI that makes real phone calls, and we had a system that would produce a confident summary of how a call went. Every summary was traceable: it pointed at a real transcript, a real call, a real record. The sourcing was never fabricated. What was fabricated was the inference. A six-second call where nobody picked up produced "she picked up and the surprise was delivered successfully." Fully traced to a genuine data trigger, and completely wrong about what that data meant.
Two things that did not fix it, in case they save you time: lowering the temperature (it was already low on the path that mattered), and adding "be honest, do not exaggerate" to the prompt. Models read those abstractly and sail straight past them. What worked was grounding each claim to something literally present in the source, plus naming the exact phrases the system was not allowed to emit unless the transcript supported them. Fabricated-confirmation rate went from about 24% to 2% across ~1,200 calls, with no measurable over-suppression.
The reason I think this lands on you specifically: in creative research the equivalent failure is not a made-up source, it is a real winner with a wrong reason attached. The ad genuinely won. The attribute your workflow traced it to is not why. Trace it back and the citation checks out, so the story is more believable than a black box would be — which is exactly what makes it expensive when it is wrong.
So: does your traceability cover the trigger, or the claim? Concretely, if a workflow says "this concept came from these three reviews and this competitor ad," can a user see the evidence for the causal step, or only the provenance of the inputs?
Not rhetorical — you have thought about this more than most launches here, or the line would not be in your tagline.
@getosmo Hi Anuj, very good 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.
@hamza_afzal_butt Every interaction with newly created content is captured on the backend, so the system can learn from the feedback users provide, think of it like swiping left or right on Tinder. Users can also access the brand brain directly, which contains all the visual style elements of the brand.
Congrats on launching! We spend way too long digging through customer reviews to find angles for ads, so pulling Trustpilot and Amazon feedback into the same workspace as ad performance sounds really handy. How does it work for smaller brands that only have a handful of reviews and a modest ad account to learn from?
@doganakbulut Hi Dogan, every brand needs to start somewhere! That's why we allow users to draw on proven ads from more established brands. On top of that, competitor reviews can also serve as a data source, think of it as leveraging the frustrations customers express about your competitors and positioning yourself as the better alternative.
Love the product and the team. From my previous launch, I met with Ozan (he is in Adomate team) and he showed me the demo of the Adomate. The most interesting part they are generate creatives by learning from your competitors. I'm planning to use it for my e-commerce business. Hopefully, in near future I can use for my apps too :)
@metehan_caliskan Thank you for your input. Initial focus is on DTC and Saas, but apps are definitely a field we further want to explore!
@s_logghe Love the transparency around the no black box approach. I'm curious how do you balance AI suggestions with preserving each brands unique voice especially when using competitor and review data?
@vishnu_kant_07 Hi Vishnu, great question. For every brand, we create a "brand brain" that captures its specific tone of voice, visual style, and product features. Users can then choose how closely they want the generated ads to follow their own brand versus staying closer to the original.
@s_logghe You mention agencies managing creative across multiple accounts as one of your target users. For those teams, is there hard isolation between client brand brains, so one client's reviews, competitor set, and performance data can never surface in another client's generated concepts?
@clement_avq Yes, 100%. None of this information is shared across brands. What agencies can replicate is the workflow itself, the process of researching data and turning those insights into new creative concepts.
@s_logghe Good to know, that's a clean line between the workflow and the underlying data! If your taste/learning settings get more sophisticated per account down the line, does the isolation guarantee extend to that too?
@clement_avq Yes, absolutely. The same isolation applies: each brand has its own Brand Brain that captures learnings and improves over time based only on that brand’s data. Nothing from one brand is ever used to influence another brand’s Brand Brain or generated concepts.
@lucas_desard Good to hear it holds at the learning layer too, that's the part that would've been easy to fudge. Should make the agency use case an easier sell!
Not gonna lie, it looks really good.
Maybe the only thing I'd like to see is to get an analysis and then prepare an ad myself.
Is it possible?
@kamil_infeld I’m not sure I fully understand your question.
If you mean whether you can first analyze the data and then create the ads yourself, then yes, that’s very much how Adomate works. You can collect and analyze data at scale, generate ad concepts and ideas in bulk based on those insights, and then refine the ones you like until they’re just right. Once you’re happy, you can launch them directly into Meta with the appropriate copy and creatives.
If I misunderstood, feel free to clarify.