Agent sessions
An AI agent's ongoing task or interaction, like a conversation or a coding session.
An agent session is like a dedicated work period for an AI. Think of it as the AI focusing on one specific job or conversation you've started.
For example, when you use GitHub Copilot to help write code, each time you ask it to suggest code or refactor something, that's part of an agent session. If you're working on a particular feature, all the interactions with Copilot related to that feature happen within that session.
An agent session represents a distinct, bounded period of interaction or task execution by an AI agent. This session maintains context, state, and history relevant to the ongoing task, allowing for coherent and progressive action. Sessions can be initiated by user prompts, automated triggers, or system commands, and they conclude upon task completion, explicit termination, or timeout.
Managing multiple sessions is crucial for productivity, enabling users to switch between different tasks or contexts without losing progress. This is particularly relevant in tools like GitHub Copilot, where distinct coding projects or feature developments can be managed as separate sessions.
The concept of agent sessions is gaining prominence as AI tools become more integrated into complex workflows. Tools like Microsoft Copilot are enabling users to manage multiple concurrent tasks, highlighting the need for clear session management in AI applications.
- GitHub Copilot app for Beginners: Managing your workMicrosoft · GitHub Blog: Copilot ·
- Microsoft Copilot
- Microsoft Copilot allows users to manage multiple GitHub Copilot AI sessions and tasks, particularly within its 'My work' pane.
- Conversation
- A conversation is a broader term for back-and-forth communication,
- Process
- A process is a general term for a series of actions. An agent session i
- Thread
- A thread typically refers to a sequence of messages in a communication p
The term 'agent session' arises from the broader field of agent-based systems and software engineering, where a 'session' denotes a period of interaction between a user and a system or between system components.
Open GitHub Copilot in your code editor. Start a new coding task, like writing a simple function. Make a few requests to Copilot for suggestions or explanations. Then, switch to a completely different file or project and start a new coding task there. Notice how Copilot maintains context for each task, effectively managing separate sessions.
The Playbook: reusable how-tos, prompts, configs and tips pulled out of the news.
- HeuristicUse the My work pane to track Copilot sessionsOrganize and track all your Copilot requests and work in one centralized location. · 5 min
- PromptRole-play a student struggling with math for AI tutor simulationSimulate a student's math struggles and uncertainty for AI tutor evaluation. · 5 min
- TemplateReusable workflow templates for healthcare teamsTeams can create and share templates for common workflows like documentation or prior authorizations. · 15 min
- TemplateTemplate for building an AI tool tailored to your workA structured approach to defining and building a custom AI tool for specific business needs. · 15 min
- TemplateEvals CSV prompt list for agent skillsA CSV file template for listing prompts used in evaluating agent skills. · 20 min
Questions people ask
- What is the difference between an agent session and a conversation?
- A conversation is the general exchange of messages. An agent session is the AI's active work and context management during that conversation or for a specific task within it.
- Can an AI have multiple sessions at once?
- Yes, advanced AI tools can manage multiple sessions simultaneously, allowing users to work on different tasks or projects concurrently without losing context.
First seen on Subvolts Aug 19, 2026. Updated Sep 15, 2026. Written by an AI model from the items above and reviewed against them weekly; report an error.