AI-generated ads, done honestlyHow we use AI to make your ads, and what we never let it do.

AI generates the pictures and the motion. From a brief we write, it produces the frames, the product scenes, the on-screen text layouts and the first cut of each variant, and it does that fast enough for a new drop every week. That speed is the point: it lets us test twelve, thirty or seventy distinct ideas a month at a price a small brand can pay, where a shoot would test one.

It also helps us resize. One idea becomes 9:16, 4:5 and 1:1 with the safe zones respected, and the captions and headlines are drafted for each placement before a person rewrites them.

Every decision that makes an ad work is a person's. The founder studied psychology, and each ad starts as a hypothesis about how your customer thinks in the three seconds they give it: the principle it leans on (loss aversion, social proof, anchoring, a problem named before the product), the hook, the offer and the proof. We write the brief, pick the idea, and read the results every week to decide what the next drop tests.

A person then checks every ad twice before it reaches you: once for the brief (does it say what we agreed, in your voice, with your prices and offer right), and once for the things AI gets wrong: hands, text, lip-sync, a claim that drifted, a safe zone cut off. Nothing ships on the strength of a prompt alone.

• The idea, the hook and the principle it tests
• The brief, your voice, your prices and your offer
• Every claim, checked against what you can prove
• The two-step check on every file before delivery
• Reading your results and planning the next drop

The same rules apply to every client in every country, because the laws on fake reviews are strict and getting stricter: the UK's Digital Markets, Competition and Consumers Act lets the CMA fine up to 10% of global turnover, and the US FTC's rule on fake reviews carries penalties per violation. We would keep these rules anyway.

• A person in an ad who is not a real customer never claims personal experience of your product. If we script a testimonial, it comes from a real, documented review you hold, and the ad says so.
• We never make fake reviews, invented customers or made-up results, for you or for ourselves. The examples on this site are labeled as examples.
• You approve every ad before it runs, and every claim in it is one you can support, as the UK CAP Code and the FTC require. If you can't, we change the ad.
• Where an ad will run in the EU and shows a realistic AI-generated person, we label it as the EU AI Act requires from August 2026. Tell us where each ad will run.
• We don't make ads for categories where AI-made claims do the most harm: supplements and health claims, income claims, crypto, gambling, weight loss or adult products.

Honestly: the law is still settling. The UK may change how it protects computer-generated work, and in the US a work needs a human author to be copyrighted. So we don't promise you full copyright, because nobody can. We promise what we can keep: a perpetual, exclusive license to run every paid-for ad anywhere, for as long as you like, plus an assignment of whatever rights do exist. It is written into our terms of service.

Because trust in AI-made content is falling while its use is rising, and a brand that is open about it stands out from one that is not. The ads are judged by your customers, in the feed, by whether they stop and buy. The method behind them is on how it works, what arrives each week is on the page about our done-for-you Meta and TikTok ad creative, and the easiest way to see the thinking is a free review of your ads.

What does that mean for a brand running our ads?

Treat every ad we deliver as AI-generated, because its frames and motion are made with AI. On TikTok, that means switching on the AI-generated content label when you upload it, or showing your own clear disclaimer in the ad. On Meta, an ordinary product ad has no box to tick, but Meta may add its own AI info label, so expect to see it on some ads.

The stricter rules start when an ad touches an election, politics or a social issue: Meta and Google then require you to declare realistic synthetic content before the ad runs. In the EU, from 2 August 2026, an ad showing a realistic AI-made person who could pass for real must say it is AI-made, and we build that label in when you tell us an ad will run there. In the US, the test is the old one: every claim must be true and provable, and no ad may pass off an AI-made person as a real customer.

This is a summary of public rules, checked on 18 October 2026, not legal advice. The rules change often, so check the platform's own page before a sensitive campaign, and ask us if you are unsure which rule covers an ad.