Visuals true to your brand, without starting from scratch.

Genie was built so teams didn't have to start from a blank canvas. You described what you wanted as a "wish", picked a style and an aspect ratio, and Genie returned a set of images that matched your brand.

I shaped the product, designed every screen and led the front-end build, while Bukola built the backend and AI behind it.

Sign up was the first moment with Genie, so I gave it a mascot. Pixel peeked and closed its eyes as people signed up, then introduced itself as their guide, which set a friendly tone before anyone had made a single image.

Generations were rarely perfect first time, so each image opened into a preview with focused actions: Edit, Select an Area, Generate similar, Vectorize and Animate, alongside Rerun, Add to Collection and Download.

Select an Area let people mark one part of an image and change only that, and Generate similar created variations with controls for how much they should differ from the original.

Style and aspect ratio sat together in the general settings beside the prompt, so people could switch between formats such as landscape and square without leaving the create flow.

Images could be saved into collections and shared with a team. Inside a collection, teammates left comments, replied, mentioned each other and attached images, so feedback stayed next to the work it was about.

Styles were how Genie learned a brand. People could upload a brand guideline as a PDF to pre-fill a style's details, or create one manually by entering the brand's colours, fonts and visual style. Teams without a style could choose from a library of curated ones.

By October 2026 Genie had over 1,500 users and early revenue. As a co-founder I made the product calls, designed them and built the front-end myself.