GitHub is rolling out a set of updates to its Copilot app that shift the AI assistant from a line-by-line code completer to something more like a project partner. The new features — project-based AI sessions, interactive canvases, and streamlined task management — are designed to help developers work through bigger problems without constantly switching context.
What the new features do
Project-based AI sessions let developers start a conversation with Copilot that stays tied to a specific project. Instead of asking isolated questions, a developer can open a session, describe the overall goal, and then ask follow-ups that build on earlier answers. The AI remembers the project context across the session.
Interactive canvases are a visual workspace where developers can sketch out ideas, drag in code snippets, and see how different pieces fit together. GitHub says the canvas updates in real time as the AI suggests changes or new code blocks.
Streamlined task management ties into the session and canvas. Developers can break a feature into smaller tasks, assign them to the AI or to themselves, and track progress without leaving the app. The task list lives inside the project session, so the AI knows what's been done and what's next.
These features aim to enhance developer productivity by reducing the back-and-forth between tools. Instead of jumping from a chat window to a code editor to a project board, everything lives in one place.
How the pieces fit together
The three features are designed to work as a unit. A developer might start a project session, describe the feature they're building, and Copilot will suggest a plan. The developer can then open an interactive canvas to visualize the architecture, drag in existing code, and ask Copilot to generate new functions. As the work progresses, the task list gets updated automatically.
GitHub hasn't said whether the features are available in the free tier or only in paid plans. The company is rolling them out gradually, so some users may see them before others.
Copilot has been mostly a chat-based tool that answers one question at a time. The new project-based approach means the AI can hold a longer conversation about a single codebase. That could save time for developers who need to refactor a module, add a new API endpoint, or debug a complex issue.
The interactive canvas is a notable addition. Visual tools for code are rare in AI assistants, and this one lets developers see relationships between components without leaving the app. It's a concrete step toward making AI feel less like a search engine and more like a collaborator.
Task management inside the session is the glue. Without it, a developer would have to remember what the AI suggested and manually track what's done. Now the app keeps the state.
The features are live now for Copilot app users. Developers can start a new project session and try the canvas and task list. GitHub says more updates are coming, but hasn't given a timeline.
