GitHub Opens Copilot Desktop App to All Subscribers
GitHub has opened its Copilot desktop application to all existing subscribers, signaling a broader shift in software development. Instead of simply assisting developers with code suggestions, the platform now focuses on AI agents that handle coding tasks while developers direct the workflow.
The Copilot desktop app first entered technical preview on May 14. However, on June 2, GitHub expanded access to Copilot Pro, Pro+, Business, and Enterprise users across Windows, macOS, and Linux. Meanwhile, Copilot Free users can join a waitlist for future access.
Unlike traditional Copilot integrations in popular development tools, the desktop app operates as a standalone platform. As a result, it removes conventional coding interfaces such as file trees, tabs, and built-in editors. Instead, it serves as a control center for AI agent sessions connected to GitHub repositories.
AI Agents Take on Development Tasks
At the core of the application is the “My Work” dashboard. This view brings together active agent sessions, pull requests, issues, CI checks, and automated tasks from connected repositories.
Additionally, each AI agent works within its own git worktree and dedicated branch. Therefore, multiple agents can operate simultaneously on the same repository without creating branch conflicts.
One of the key features is Agent Merge. Once enabled, it reviews pull requests, responds to review feedback, fixes failed CI checks, and completes merges when predefined conditions are satisfied. Consequently, Copilot becomes more than a coding assistant and takes part in the review and integration process.
Another major addition is Canvases. These shared workspaces allow developers and AI agents to view, edit, and update structured content together in real time. For example, a canvas can contain project plans, code changes, terminal sessions, deployment dashboards, or release checklists.
GitHub describes the feature as the start of a new “agent experience” model, where humans and AI agents collaborate through shared interfaces.
New Features Support Agentic Development
The launch coincides with GitHub’s transition to AI Credits, a usage-based billing model that took effect on June 1. Because agent-driven workflows require significantly more computing resources, the new system links costs to token consumption rather than fixed subscription fees.
GitHub had already paused new individual plan sign-ups in April due to the growing compute demands of agentic development. Now, the expanded desktop app aligns with that broader strategy.
Furthermore, the latest release introduces several new capabilities. These include voice input powered by on-device speech-to-text, cloud-hosted agent sessions, scheduled automations, and an integrated browser that allows AI agents to verify their own interface changes. Together, these additions further extend the role of AI agents within the software development process.
