The GitHub Copilot application runs multiple AI agents in parallel on the same project without interfering with one another. Users can start another task without waiting for one to finish; each session’s settings and workflow are kept separate. A session refers to the task given to Copilot from start to finish. The task name and progress status can be seen on the cards in the Sessions view.
Since each agent session in the application runs in its own Git worktree, tasks can be carried out simultaneously. Each session also maintains its own context; when users switch between them, the task continues from where it left off, and there is no need to explain the operations again. This allows users to focus on reviewing results and making decisions rather than monitoring the agents.
For example, in the tailspin-toys repository, tasks such as adding funded sort, conducting an accessibility review, and running tests can be started in three separate sessions. Progress can be tracked through the Sessions view. GitHub recommends starting two small tasks at the same time in the application to try out the feature.
Why it matters
This approach directly concerns users who want to run workflows that would otherwise wait for one another separately during software development. Isolating tasks in Git worktrees allows changes within the same project to progress without interfering with one another, while preserving session context eliminates the need to rebuild the workflow each time users switch between sessions. As a result, the way Copilot is used moves closer to a workflow in which different tasks are started and their results compared, rather than having a single task completed sequentially. The Sessions view provides a central view for tracking this workflow; however, the responsibility for reviewing the results of numerous simultaneous tasks and deciding which changes to merge remains with the user.
Background
GitHub, FikirPilot’s archive is no stranger to this name: we published a news article mentioning it in the past 90 days; that article is dated August 27, 2026.
Term: agent
An artificial intelligence agent is software that calls tools and carries out multistep tasks to achieve a goal, rather than producing a single response.