Build a website in your AI Workspace.

Start with the page or app you want to use. DevShot brings the AI conversation, working files and tools together so you can build, inspect and test in the same project.

Start with a concrete request

Describe the audience, the job the page needs to do and the interaction that must work. For example:

Build a small project tracker called Fieldnotes. Use three project cards and an Add task button. Let me add a task, mark it complete and check the result in the browser. Keep the project local.

This gives the work a clear acceptance check: the button must add a task and the completion control must change its state.

Work through the result

1. Open a project in DevShot and submit your request.
2. Follow the AI's progress in the conversation.
3. Open the app and use its controls. Check a narrow screen as well as a desktop screen.
4. Use the editor to inspect the source and the terminal to run the project's tests.
5. Describe the next change against the result you can see.

Keep changes reviewable

Ask for a small change at a time. State the expected behaviour: “Keep the added task after a reload” is more useful than “Make it better.” Verify links, forms, empty states and keyboard interaction before publishing.

Building a project and publishing it are separate actions. You can inspect and test locally before deciding to publish.

Continue the project later

Keep the files and conversation with the project. Check its backup status before ending your session. Pricing explains how workspace running time is billed across projects.

Open your AI Workspace or learn how workspaces and projects fit together.