Multi Agent Collaboration Gets Persistent Compute in Bedrock AgentCore
Amazon Web Services has extended Amazon Bedrock AgentCore with runtime instances, a new compute option that gives AI agents persistent infrastructure purpose-built for complex long-running workflows and multi-agent coordination. By Matt Saunders
Amazon Web Services has introduced runtime instances in the Bedrock AgentCore, adding a second compute option that runs agents on managed Amazon EC2 in a customer account. This new feature keeps the existing AgentCore APIs, identity and observability model, providing persistent EC2-backed sessions for agents to run for up to fourteen days.
These sessions support shared file systems, GPU accelerated instance types, and compatibility with Python and container images. Multiple agents can collaborate on the same host through a shared session directory, eliminating the need for each agent to call each other's APIs for handoffs. The runtime instances are designed to support multi-agent collaboration, complementing the eight-hour microVM ceiling that AgentCore launched with.
AWS positions runtime instances as a complementary compute option, encouraging teams to use a mixed topology approach where orchestrator agents on microVMs dispatch long-running work to workers on instances. A cost analysis suggests that runtime instances are financially viable when agents have a minimum of 24% sustained CPU utilization, factoring in EC2 commitment discounts.
The introduction of runtime instances aligns with the rise of agent orchestration platforms and offers a hybrid pattern for development teams.
Written by urgent.news from InfoQ's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
