Create with AI Part II & the unveiling of Clicky Create

Yesterday we filled Clicky HQ in Manchester for the 2nd instalment of Create with AI, and once again the room was at capacity.

Real or AI?

We opened with a game. Eight images on screen, one question each time: real or AI? The early rounds got a decent hit rate. By the halfway point the room had stopped putting hands up with much conviction, and by the end nobody was confident about anything.

To make the point properly, we went back to 2021 and a DALL-E generation of "Heisenberg eating instant noodles with a cat". It is a lovely piece of nonsense, and it is four years old. The distance between that image and the ones we had just been arguing over is the whole reason for the event.

Everyone can make one image. Almost nobody can make a thousand.

This is the gap, and it is where most brands quietly give up.

Open Gemini or ChatGPT and you will have something good within seconds. No skill barrier, no budget, no brief. Everyone in the room had already done it.

Now try to produce a thousand. The product comes back wrong. The brand drifts between generations. Hands and faces fail. Nothing is repeatable, so nothing can be signed off, and the whole exercise stalls at the point where it would actually start earning its keep.

That is the problem we set out to solve.

Why Gemini, honestly

We spent a slide on the model landscape, with the caveat that it changes roughly every six weeks.

GPT Image 2 currently sits top of the Artificial Analysis Image Arena, by the largest margin the leaderboard has recorded, and has the strongest raw aesthetic of the three. Midjourney v7 still has the best pure look and feel, but it is the weakest on control and repeatability, and it is hard work holding a real product steady.

We use Nano Banana Pro, Gemini 3 Pro Image, because it gives us identity lock across generations, native 4K with multi-image blending, and the strongest grounded reasoning of the lot. We are not picking the highest-scoring model. We are picking the one that holds a product and a brand steady across a thousand generations.

The obvious follow-up came up, as it always does: why not just use Gemini yourself? Consumer accounts carry daily limits and a visible watermark, and are not built for commercial production. The commercial-grade API gives effectively unlimited usage, no watermark, cleared commercial use, and output up to 8K, which is print and paid ready.

Clicky Create, live on stage

Then we showed the thing itself, for the first time anywhere.

Powered by Gemini, Clicky Create is an image platform that learns a brand and then holds it steady at scale. It applies the lockup correctly every time rather than approximately. It learns the product ranges, so the right product from the right range comes back rendered as it actually is. It places the correct families into the correct scenes at the correct scale. And every failure feeds back in, so the same mistake stops recurring.

The demo ran in seven steps. Add the client by website address and nothing else. The platform reads the site, works out the product ranges and imports the imagery. Pick the range, a country kitchen in our case. Pick the family. Pick the season. Choose your room angles, then your cut sizes, Reel, square, whatever the placement needs. Press process.

The part that generates the most interest is the bit in the middle. Assets are produced against the brief, judged against our own criteria, and anything that fails goes straight back through the model to try again. Only verified output reaches you. That step is the difference between interesting output and usable output.

Then, at the very end, we sent a finished still through Google Veo and watched it move. Same source image, same brand, same product, now a social video asset from the same catalogue.

Why any of this matters

We are a digital marketing agency, not a lab. The reason for the technology is better campaign results.

Imagery with people in it outperforms imagery without it. A holiday cottage or a new kitchen shot empty simply does not work as hard in a paid placement, and adding people to imagery that has none is the single most common request we get. That is where Create starts. The approach extends a long way past it, but that is where the demand sits today.

The legal picture

Sergio Ferreira, Intellectual Property Solicitor at Knights, closed the session with the part most brands worry about: usage rights, who actually owns a generated image, and where the real risk sits for a brand using this commercially.

We put him after the demo on purpose. See the capability first, then hear the legal position. Our thanks to Sergio for a clear and reassuring half hour, and for taking questions alongside us at the end.

The short version of the night: the capability is here, and the legal picture is clearer than most people assume. There is not much reason left to wait.

Want to try it?

Clicky Create is in beta, running as a standalone app while we get it right, and it will become the first creative tool inside Clicky One, our client platform. It runs on credits.

We are looking for one or two clients to trial it properly. If that might be you, get in touch.