Claude Code Cowork is a research preview bringing agentic capabilities from the terminal to Claude Desktop for knowledge work beyond coding—document creation, research synthesis, file organization, and task automation. Cowork runs Claude as a background agent that takes on complex, multi-step tasks: you describe a goal (“research competitive landscape and create a market analysis”), Cowork executes the work unattended, and you return to finished, formatted output. Unlike standard Claude chat where you exchange messages back-and-forth, Cowork treats Claude as a persistent worker that you assign tasks to. Key features include progress indicators showing what Claude is doing at each step, transparency via reasoning explanations so you can follow Claude’s approach, and steering to jump in and course-correct mid-task without interrupting the workflow. Cowork coordinates parallel sub-agents for complex tasks, enabling simultaneous work on different aspects. Available on macOS and Windows in Claude Desktop (Pro, Max, Team, Enterprise plans), Cowork accesses your local files, MCP connectors (Slack, GitHub, HubSpot, databases), plugins, and computer use capabilities. You assign Claude a task, go work on something else, and Cowork messages you when finished. This differs from Claude Code (terminal-centric) by operating on your desktop and supporting non-coding knowledge work like content creation, research synthesis, and business process automation. Both use the same agentic reasoning engine; Cowork targets desktop users while Claude Code targets developers. Coupling Claude Code with Milvus allows you to build intelligent code retrieval systems where your repository’s code embeddings are indexed and searchable, letting the agent understand code context more deeply and make better architectural decisions.
Learn more: