Amazon Bedrock Introduces Multi-Agent Collaboration Capability in Preview
Amazon Web Services (AWS) has launched a new multi-agent collaboration capability for its Amazon Bedrock service, available in preview. This feature enables multiple AI agents to work together on complex, multi-step tasks.

Amazon Web Services (AWS) has released a preview of its new multi-agent collaboration capability for the Amazon Bedrock service. This new functionality allows multiple specialized AI agents to coordinate and work together to complete complex, multi-step tasks.
For advanced applications where a single AI agent is insufficient, multiple agents with distinct areas of expertise can now be utilized. This feature significantly simplifies inter-agent communication, task delegation, and the consolidation of outputs. Previously, developers using open-source solutions had to manually manage technical challenges related to agent orchestration, memory management, and session handling.
The managed multi-agent collaboration capability on Amazon Bedrock employs a "supervisor agent" that breaks down an overall request into smaller components and delegates them to appropriate specialist agents. For instance, an investment advisory system could comprise an agent for financial data analysis, another for research, a third for forecasting, and a fourth for investment recommendations. Similarly, in retail operations, agents could handle demand forecasting, inventory management, and supply chain coordination.
AWS states that this capability enhances task success rates, accuracy, and productivity by enabling seamless collaboration among agents. The general availability of the multi-agent collaboration feature has been announced for a later date.