AI can generate a lot. It can’t decide what actually works.

This is my Human by Design entry which is a full product promo teaser, built entirely with envato’s AI toolkit, split into 3 parts in one video:

The hook: setting up the problem before you even see the product.

The build; every prompt, edit, regenerate, and dead end (yes, including the ones that didn’t work)

The result is a final cut polished enough for a real client to post today.

The moment that mattered most: mid-shoot, the AI video generation got flagged and blocked outright. No amount of re-prompting fixed it. So I stepped in and redirected the entire visual direction myself the shots, the pacing, the story beat it was supposed to hit.

AI can execute. It can’t make that call. That’s the difference this challenge is really asking about not what AI can produce, but where a human still has to step in and shape it.

Made with Envato’s Gen AI + stock tools.

Soar for the laurels

You're divinely backed up

Congratulations 🔥🔥🔥

One day I want to open a bagel shop. For now, I got to prototype the whole brand instead, logo, packaging, poster, video using Envato's gen AI tools. It's a surprisingly solid all-in-one kit for image and video generation.

My Workflow

- Sketched the logo on paper first, pulling inspiration from Envato's geometric patterns, I go for a monochromatic, warm, geometric logo.

- Built the final logo in Illustrator, then handed it to AI with clear sketches and direction.

- Used AI to rapid-generate mockups across t-shirts, packaging, and posters.

- Iterated through a brand animation, testing multiple storyboard approaches before landing on the final version. - Generate music via Envato AI then edit in After Effects.

My Human by Design Moment My first animation storyboard for the logo was low-fidelity, so the AI output felt random. I assumed the fix was more detail, so I redid it with the final logo and tighter instructions. AI overcorrected. It added motion and elements I never asked for, and the result was more confusing, not less.

I made a different call. Instead of adding more, I stripped the storyboard down to just two frames: a start and an end. No extra instruction for AI to misread, no room for it to improvise. Designers has design in mind, our goal is to visualise that vision. AI helps along the way, but only when I gave it clarity. Simpler instruction with examples what finally got a clean, consistent animation out of it, one I could adapt across other bagel without starting over. Curious on how other make the AI works? Youtube: https://youtu.be/-HQX598KRK0

I love the color palette and grading of the photos too! They look so good!