AWS announced the new Amazon Bedrock AgentCore runtime, a managed compute layer for production agents, adding on-demand memory allocation with reclamation of cold memory and snapshot-based cold starts that stay consistent regardless of image size. A companion AWS post demonstrates migrating a multi-model healthcare agent from ECS/Fargate to the runtime with no agent-logic changes. The updates target teams running long-running, bursty agents where peak-memory billing and variable cold starts inflate cost and latency.
AWS published a blog post on optimizing production agents with Amazon Bedrock AgentCore Observability, focusing on performance issues like slow response times and unbounded memory growth. This is part 2 of a series, following part 1 on debugging infinite loops and tool invocation errors.
AWS published a blog post on optimizing production agents with Amazon Bedrock AgentCore Observability, focusing on performance issues like slow response times and unbounded memory growth. This is part 2 of a series, following part 1 on debugging infinite loops and tool invocation errors.
An AWS blog post argues that agents working in notebooks fail in production, where teams inherit operational work unrelated to the agent's reasoning. It maps ten such burdens and positions Amazon Bedrock AgentCore, plus Bedrock Guardrails, as the migration path.
An AWS blog post argues that agents working in notebooks fail in production, where teams inherit operational work unrelated to the agent's reasoning. It maps ten such burdens and positions Amazon Bedrock AgentCore, plus Bedrock Guardrails, as the migration path.
AWS published guidance on asynchronous invocation patterns for Amazon Bedrock AgentCore agents in serverless pipelines, which remove idle compute costs while agents process requests. The post uses a real-estate financing document validation example to illustrate the pattern.
AWS published guidance on asynchronous invocation patterns for Amazon Bedrock AgentCore agents in serverless pipelines, which remove idle compute costs while agents process requests. The post uses a real-estate financing document validation example to illustrate the pattern.
An agentic platform AWS is building with Wood Mackenzie on Amazon Bedrock AgentCore, addressing the gap between agent prototypes and production deployments. It standardizes concurrency, session isolation, identity, state, scaling, and guardrails so the energy-research firm doesn't rebuild this layer per agent. LLM-based agents run on real traffic with monitoring to verify correct behavior.
An agentic platform AWS is building with Wood Mackenzie on Amazon Bedrock AgentCore, addressing the gap between agent prototypes and production deployments. It standardizes concurrency, session isolation, identity, state, scaling, and guardrails so the energy-research firm doesn't rebuild this layer per agent. LLM-based agents run on real traffic with monitoring to verify correct behavior.
An AWS post walks through deploying a three-agent music production pipeline on Amazon Bedrock AgentCore Runtime Instances, the new EC2-based compute option for long-running agent workflows. Three Strands Agents (composition, delivery, compliance) colocate on one GPU instance via a shared session ID, render a track with ACE-Step, master it, and screen it against a back catalog. The architecture demonstrates patterns AWS says extend to 3D rendering, simulation, inference, and media processing workloads.
An AWS post walks through deploying a three-agent music production pipeline on Amazon Bedrock AgentCore Runtime Instances, the new EC2-based compute option for long-running agent workflows. Three Strands Agents (composition, delivery, compliance) colocate on one GPU instance via a shared session ID, render a track with ACE-Step, master it, and screen it against a back catalog. The architecture demonstrates patterns AWS says extend to 3D rendering, simulation, inference, and media processing workloads.