AWS brings multi-agent orchestration to Bedrock

AWS is doubling down on AI agents with the announcement of multi-agent capabilities on its Amazon Bedrock platform.

During his keynote at the AWS re:Invent conference, AWS CEO Matt Garman said that customers building agents in Bedrock wanted a means to make agents work together.

“While a single agent can be useful, more complex tasks, such as performing financial analysis across hundreds or thousands of different variables, may require a large number of agents with their own specializations,” Garman said. “However, building a system that can coordinate multiple agents, share context with each other, and dynamically direct different tasks to the right agent requires specialized tools and expertise in generative AI that many companies don’t have available.”

The new capabilities allow companies using Bedrock to create agent workflows to build AI agents and establish their entire agent ecosystem. This includes the ability to create orchestration agents to manage multiple agents and workflows that require multiple steps.

AWS’s emphasis on agent collaboration differs from the approach taken by Microsoft in its announcements about multi-agent support two weeks ago. AWS’s approach is based on its extensive experience with microservices. In an interview with VentureBeat, Swami Sivasubramanian, Vice President, AI and Data, AWS said the company gained key insights from the development of its Q development agent, which it says is best-in-class on the SWE benchmark, which is based on tasks real world engineering. That background shaped the tools being introduced today, he said. This production-ready approach allows companies to move from prototype to deployment faster, he said, with orchestration capabilities that streamline workflows, manage state sharing and dynamically assign tasks between specialized agents. These features differ from AWS competitors like Microsofthe said, agent tools prioritize broader frameworks, but don’t have the same level of focus on orchestration and deployment readiness.

“Using multi-agent collaboration in Amazon Bedrock, customers can achieve more accurate results by creating and assigning specialized agents to specific steps in a project and speeding up tasks by orchestrating multiple agents working in parallel,” AWS said.

Customers build their specialized agents in Bedrock and then make a supervisor or orchestrator agent to help manage the other agents. AWS said that the supervising agent “is responsible for coordination, such as dividing and directing tasks to the appropriate agents, giving specific agents access to the information they need to complete their work, and determining which actions can be processed in parallel”.

After the other agents finish their tasks, the orchestrator agent aggregates all this information. Garman pointed out Moody’sthe credit rating agency and one of its customers that first used the multi-agent capability. Moody’s created a number of agents for its risk analysis workflow. The company created agents that analyzed macroeconomic trends or looked at corporate risks, which were able to build more accurate risk assessments.

AWS first introduced agent capabilities in 2023 with the release of Agents on Bedrockwhich gave companies a way to start building agents. During its re:Invent event this week, AWS customers such as PagerDuty and GitLab announced new agents that allow users to develop their applications on their platforms and use AWS agents to improve their workflows.

Multiple agents are becoming the norm

Companies are turning to agents to help simplify their workflows, so having an agent ecosystem with multiple AI agents is quickly becoming popular, and service providers have taken notice. Microsoft provides a library of agents for Copilot users (and has accumulated one of the ecosystems of larger agents). Service Now also has a set of AI agentsmaking his orchestrating agent a big selling point. even OpenAI understands the growing attraction of AI agents with their Swarm AI Agent Framework.

However, companies also need to monitor the proliferation of AI agents and ensure that the agents actually do their job for the task. The orchestration layer, often consisting of an orchestrator agent, monitors the progress of the task and triggers which agent will start working next.

AWS’ focus is on enabling businesses to build the complex multi-agent workflows they want and the orchestration layer they need. Other providers, such as ServiceNow and Salesforcethey give customers access to agents, which can be customized to their needs, and then offer orchestrator agents built by the company.

Managing the expansion of AI agents is the next big step in the water wars, and the space is heating up.