Collaborative AI Agent Framework CoAgent Launches: Multiple AI Agents Autonomously Divide Labor on Complex Task Chains

German AI startup Sapienz released the CoAgent collaborative agent framework on November 2. The framework allows users to direct multiple specialized AI agents through a single interface. One handles information retrieval, another performs data analysis, and a third drafts documents. The agents autonomously negotiate task division and share work context among themselves.

Unlike existing single-agent systems, CoAgent includes a built-in negotiation protocol between agents. When a task spans multiple specialized domains, agents automatically assess their capability boundaries and propose optimal division plans. In early tests, CoAgent improved efficiency on market research report generation by 3.2x compared to single-agent approaches. The framework is released as open source and supports integration with GPT-4o, Claude, and Gemini.