Bottom line: Google introduces subagents for Gemini CLI: specialized AI agents that handle complex tasks in parallel and make the main session more efficient.
Google has introduced subagents for Gemini CLI, specialized AI experts that handle complex and recurring tasks in parallel to the main session, thereby making it faster and more efficient.
Subagents are specialized AI experts that operate independently alongside the Gemini CLI main session. When a large or complex task arises, Gemini CLI acts as a strategic orchestrator and distributes specific subtasks to the most suitable subagents.
Each subagent operates independently and has its own tools, MCP servers, system instructions, and its own context window. This enables the main session to remain fast, clear, and focused on the overall vision, while specialized subagents handle the intermediate tasks.
The new feature enables large amounts of work to be efficiently distributed across multiple agents, which is particularly advantageous for complex projects. Subagents work in parallel, which accelerates processes and increases overall productivity.