Build and share interactive prototypes in Codex

TL;DR: The OpenAI product design plugin lets designers go from concept to a shareable interactive prototype entirely within ChatGPT, with one-click export to Figma and publishing as a team-accessible website. ## Overview The OpenAI product design plugin changes how product designers work. It turns the most challenging part of design—from idea to shareable prototype—into a streamlined workflow, enabling companies to build better products faster. Below is a walkthrough showing the full process: designing a new calendar feature inside ChatGPT. ## Clarify Direction & Ideation First, the model clarifies the direction by asking a few questions. We answer them and provide a reference file as a visual brief. Once the model has a clear understanding, it moves into ideation. The model generates three different visual directions. We can review each, compare options, and decide which prototype to build. All three are good, but we choose option one and proceed. ## Auto-Build an Interactive Prototype Based on the selected image, the model starts building an interactive prototype with code. It generates the supporting images and assets needed for the experience, tests the prototype itself, and then hands it over to us. It checks different screen sizes and even compares the reference image against the live prototype to ensure everything looks good. ## Testing & Iteration Now we can test it ourselves. Open the prototype in full screen and start clicking. Interact with the app—toggle features, scroll around—to get a real feel for how this calendar feature would work. If we want to refine something, we can point to a specific part of the prototype and request changes. Looks good so far—let’s continue. ## Share with the Team: Figma Artifacts Next, we share this idea with the team. One way is to create Figma artifacts. This sends the work into an existing Figma canvas, or we can generate a new one. Once done, we open the Figma file. We can see the new “Agenda co‑pilot” page. It doesn’t just send a screenshot—it brings the full context of the feature into Figma so the team can continue editing and refining. Prototype screenshots, user story context, and even review notes for the feature are placed directly in the file. ## Final Share: “Sites” Publishing The last step is sharing with the whole team. We use the new “Sites” feature. Once opened, the original prototype now appears as an interactive website that the entire team can access. They can browse, click on different elements, and experience firsthand how the feature behaves inside ChatGPT. ## Summary What used to be a drawn‑out process is now consolidated in one place, helping you build and deliver better products faster. All thanks to the OpenAI product design plugin. Source: Build and share interactive prototypes in Codex - OpenAI (https://www.youtube.com/watch?v=_9ImjmzAyus)