one place to make images and video with AI

Image generation had become a core tool for our firm, but it introduced new challenges. The company's Enterprise account capped everyone at three images a day, the best outside tool was hard to log into, and people were buying their own subscriptions with no view of cost and no control over what models client material went into. Meanwhile the models changed every few weeks. The firm needed one signed-in place to the firm to securely and confidently generate AI assets.

We built an internal app that created a unified platform for creating images and videos. It leverages the firm's own AI accounts, with every image and video tagged to a project code and saved to a shared gallery, and new models added as they arrived.

I led the design of the experience as it grew, and developed features alongside Claude Code and Codex. People typically arrive with a task, not a model, so the tool is organised around six jobs, from making an image to extending a video, each with its own simple workspace, model picker, and controls and constraints specific to the model. Every result, yours or a colleague's, opens in a viewer that shows how it was made and offers tools to remix, extend, and more, which makes the gallery a place to learn as well as create.

I also designed and built Storyboard mode for the harder problem of consistency. You describe a story, a planner breaks it into shots, and characters, places, and a visual style are defined once and carried through every frame, each of which can be regenerated, edited, and animated.

Imagine is how the firm makes 75% of our generated imagery and video now, from marketing visuals to wall-sized event graphics to the office's own tutorial videos, with the cost visible and billable by project. The structure has held up as the models have kept changing. New ones have been added steadily since, and each arrives as one more option inside a task people already understand.