Marketers Can Build Campaign Studios From Briefs, Brand Books, and 3D Models
OpenAI’s marketing team argues that ChatGPT Work can turn a campaign brief, brand book, product model and wireframe into a custom production tool rather than a generic image generator. In a fictional NOVA Electronics launch, a non-coding presenter built a Composition Studio for adapting headphone creative across campaign formats, alongside a product website, using supplied brand and design materials to define both the output and the workflow.

A launch brief can become a reusable production interface
A marketing team can turn its brand rules, product assets, channel formats, and preferred ways of working into a custom creative-production tool. In the fictional NOVA Electronics launch, the stated problem is familiar: creative teams are stretched thin, budgets are tight, and people are committed to other projects. The proposed response is a purpose-built application for the NOVA ONE headphone campaign.
The resulting “NOVA ONE Composition Studio” is a web app for creating and refining marketing layouts across out-of-home, social, print, and other placements. It is not presented as a generic image generator. It is configured around the campaign’s particular product, visual identity, asset inventory, and layout requirements.
The presenter, who says he does not code, built the tool alongside the campaign art and a NOVA ONE product website. The practical proposition is that a marketer can specify the system creative work needs to operate within, then use that system repeatedly to make and assess assets.
The source’s strongest evidence is in the inputs provided to define the application. The request includes a NOVA brand-book PDF, an .fbx 3D file for the headphones, a list of advertising inventory types, and a hand-drawn wireframe for the tool itself. Together, those materials establish both the visual rules and the working environment: what the product should look like, which outputs are needed, and how a user should manipulate a composition.
The inputs specify both the campaign and the working method
The tool request asks for a web app that can create visual designs for out-of-home, social-media, and print assets, with a 3D preview sized to each advertising inventory’s dimensions. The app is therefore scoped to a production problem: adapting a campaign across distinct placements.
The attached headphone model is meant to be interactive rather than a fixed product image. The prompt asks the user to reposition the NOVA ONE model by clicking and dragging, rotate it around its center with shift-drag, and scale it with scrolling. It also requests settings-panel controls, correction for original-model scale problems, and coordinate adjustments that would otherwise place the scaled product outside the camera view.
The requirements extend to the visual treatment of the model. They specify medium-gray headphones with white ear cushions, plus lighting controls for orbital position and color. The completed interface shows controls for camera orbit, field of view, zoom, position, scale, background color, and product color. Its canvas separates type and logos, 3D headphones, and background-color layers.
The brand book is not incidental reference material. The presenter says it was attached because the tool needed to match NOVA’s style guide; the completion message says the application’s styling followed that PDF. The wireframe adds a second kind of instruction. It sketches an “NOVA asset design tool” with canvas, list, preview, settings, and save areas, defining how the presenter expected to work—not merely what image to produce.
ChatGPT’s completion message reports that it worked for seven minutes and 13 seconds and built the Composition Studio locally. It lists interactive 3D headphones, independently draggable text layers, lighting and background controls, per-format saving, fixed panels, and full-resolution PNG export. It also notes that publishing would require uploading the project and supplied model assets to a private Sites repository.
Production controls matter because assets need revision, not just generation
The Composition Studio is presented as a layout environment where a user can move copy, adjust it with control points, bring in the actual headphone product, and scale the product within a composition. That combination of editable layers, product geometry, and format selection makes the interface usable for changes after an initial design exists.
The available formats visible in the tool range from standard and animated digital billboards to airport placements including baggage-claim takeovers, terminal screens, and digital networks. A team can therefore work through placements as defined campaign outputs rather than treating every asset as the same canvas.
Generation is part of that workflow, but it is not positioned as the final decision-maker. The app includes a generation library with 39 generated images, seven of them favorited. Individual images can be opened, favorited, or downloaded.
The examples shown vary product color, backdrop, and copy. One features a blue-and-white headphone on light blue; another uses pink-and-white headphones against dark blue with the line, “Our sound. Your personality.” The presenter describes the value as accelerating ideation: producing many visuals for the team and ultimately for leadership to review.
Human judgment remains in the choice of source materials, the definition of formats and controls, and the review of generated work. The source does not suggest that those decisions disappear. It shows a marketer choosing a brand system, product model, wireframe, and desired interaction patterns before using generation to produce options within that setup.
The same creative direction can extend to a customer-facing site
The NOVA materials also support a separate website build within the same project. The request supplies a strategy PDF, an image pack, and wireframe mocks, and asks for a fixed top navigation with the NOVA logo. It directs the site to use the strategy document for its main sections and copy, and references mocks for the cover, customer-profile, color, sound, and film sections.
The creative direction is specific: the site should be fun and minimal, use a white background and more colorful accents than typical NOVA work, while retaining the font system from the brand guide. The presenter also provides an art pack of images he had created and liked.
The resulting site uses navigation for Colors, Sound, Films, and “Build yours.” Its opening copy describes NOVA ONE as “Precision sound. Finished your way,” offers eight finishes, and lists a starting price of $99. A color selector displays a finish such as Sky Blue and an add-to-cart action. The sound section includes a rotatable 3D headphone model and product copy about balanced sound, noise-canceling fidelity, wide dynamic range, and soft-touch comfort.
The presenter says the wireframe was turned into the resulting style, and says the tool, artwork, and website were built “in basically a day.” The demonstrated outputs are concrete: format-specific layouts, a reviewable generation library, and a launch-site prototype shaped by supplied strategy, brand, product, and design materials.