When people hear “AI-generated ad,” they often imagine something that looks cheap, strange, or obviously fake. Like weird hands, unnatural movement or a robotic voice. Slop content exists. But it no longer represents the limits of current technology. Last month, we made and ran more than 100 fully AI-generated video ads for our clients on Meta Ads with the majority performing at par or better than their existing "traditional" creatives. **Fully AI-generated does not mean fully AI-directed** When I say “fully AI-generated,” I mean AI produced most or all of the media inside the ad: images, video, voices, and music. But while the production can be fully AI-generated, the thinking behind it remains human. Someone still needs to decide who the ad is for, which problem it should address, what it should say, and how it should make the viewer feel. Someone needs to brief the models, evaluate the results, and decide what is ready to publish. **The best AI ads start before we generate anything** A good ad still needs a relevant problem, a clear promise, a strong angle, and a reason to believe. Before generating anything, we need to know: * Who are we trying to reach? * What problem are they experiencing? * What do they believe about the available solutions? * What would make this product relevant to them? * What objections might stop them from buying? * Why should they trust this brand? AI has not made those questions less important. By making production faster, it has made weak thinking easier to expose. **Cheaper production does not remove the need for taste** Some people imagine businesses generating endless ads with almost no human involvement. We are not there today. I am also not convinced that removing people from the creative process should be the goal. Given the same generic information, AI tends to produce familiar outputs. The hooks sound alike. The scripts follow th...

Solo founder and Founders team We're live on Product Hunt today, so I'll be straight about what this is and what it isn't. No traction numbers to brag about yet - we're early access and that's the honest state of it. The insight the whole product sits on I've run paid social for a while, and every time an account plateaued my first instinct was to blame targeting. It was almost never targeting. It was that we ran the same three creatives for six weeks, because a new UGC batch costs one to two thousand dollars and takes two weeks to come back. So you ration. You pick the three safest angles, run them until they fatigue, then scramble. Creative isn't a production problem. It's a search problem. You're not trying to make one great ad you're trying to find which promise your market responds to. That's a search across fifteen angles, and you can't search a space of fifteen with a budget for three. What it does You paste a product URL. It fetches the product page, extracts the product details and brand tone from the site itself, writes the hooks and ad copy, generates fifteen image and video creatives UGC-style reels, statics, comparisons, routine embeds — and pushes them to Meta ready to publish.About thirty seconds. No brief, no editor, no shoot. The demo video is a real run, not a mockup: a forty-two dollar vitamin C serum from a skincare brand, URL to fifteen publish-ready creatives in under a minute. The build lesson that cost me months V1 asked for brand colors, product photos, a creative brief, and format selection. Six steps. Almost nobody finished. I assumed positioning was broken and rewrote the landing page twice. Then I actually watched people use it. Everyone died at the same step - the brief. Obvious in hindsight: writing a creative brief is the exact job they came to outsource. I'd built a tool that made you do the hard part before it did the easy part...

I’ve been testing some AI video ad creation tools for our Meta advertisement campaigns for about two months now. there’s no denying that the high-volume type tools (the URL-to-video type of programs which generating UGC-like videos in bulk at a rapid speed) do deliver when numbers are the main priority. However I find I need to clean up a lot after generation to discard videos that dont fit either due to the brand voice, awkward tone or lip-sync issues, and to find out the angles of the videos on my own. Another direction I tried is looks more like end-to-end creative focused tools as opposed to assembly-line like mass content creation. It does video and static creation but seems to put a larger degree of attention on brand guidelines and consistency which leads to fewer unwanted items in its resulting output but those that still remain are less obvious and require extra care to eradicate before deploying. I’m mostly curious whether there strategies to fix the weaknesses of both or if this’s the “doesn’t work for you = isn’t meant for you” case meaning the difficulties I face are of the very nature of these tools and I should either accept it or not use at all. I do hope there’s a way to streamline the creation process furtherly. In case it makes it complicated to figure out what tools I was talking about describing the categories they belong to, it’s Creatify which I described as a bulk-creation tool and Omneky for the end-to-end category of tools. The residual filtering and strategy work is something I’m trying to shorten eventually.