GitHub Copilot Brings Mockups to Life by Generating Code from Images

GitHub is rolling out some game-changing updates for GitHub Copilot, and one of the coolest new features might just revolutionize how we turn design concepts into code.

With Vision for Copilot, developers and designers can now upload screenshots, photos, or diagrams directly into Copilot, and the tool will generate the code for the interface and even alt text. It’s a huge step forward in making the design-to-code process more seamless.

Here’s how it works: Let’s say you’re on a marketing or design team, and you want to make changes to a webpage. Instead of having to write out a detailed text description, all you have to do is upload a screenshot showing exactly what you want.

Copilot takes that image and instantly generates the code needed to implement those changes. No need to explain anything in words – Copilot just gets it from the image.

This feature isn’t entirely new—it’s been available as an extension in VS Code since October. But now, GitHub is bringing it fully into Copilot Chat, and you won’t need an API key or third-party extensions anymore. It’s all integrated into the core experience, making it easier than ever to work with Copilot.

Along with Vision for Copilot, GitHub is also introducing a smart feature called Next Edit Suggestions. Previously, Copilot suggested code based on where your cursor was.

But now, it looks at recent changes in the code to predict what you’re likely to do next. It’s like Copilot is starting to read the room (or the code) and is getting better at predicting what comes next. You can easily accept the suggestions by hitting Tab or reject them with Esc.

But that’s not all. GitHub is also focusing on what it calls agentic capabilities—basically, making Copilot smarter and more autonomous. With Copilot Edits, developers can use natural language prompts to make changes across multiple files.

Instead of manually selecting which files to update, Copilot does that for you, figuring out which files are relevant to the changes you want to make. The goal is to eventually have Copilot handle even more of the process with minimal input from the developer, making the whole workflow smoother.

Looking ahead, GitHub is working on something called Project Padawan, which aims to create a more autonomous software engineering agent. This would allow Copilot to handle entire tasks, not just small edits, all while staying under the direction of the developer.

While it’s still in the early stages, it’s clear that GitHub is working toward making Copilot more of an all-in-one assistant, capable of handling more complex tasks with less oversight.

The future of GitHub Copilot is all about making the development process faster, more intuitive, and more efficient. As AI takes on more of the heavy lifting, developers will be able to focus on the creative and strategic aspects of their work, while Copilot handles the rest.

GitHub is eager to get feedback from the community and work with partners to refine these features, so we’re just getting started!

Go to GitHub Copilot

Louise North

Louise is a staff writer for WebDesignerDepot. She lives in Colorado, is a mom to two dogs, and when she’s not writing she likes hiking and volunteering.