Amazon Bedrock Agents Classic is in maintenance mode. As of July 30, 2026, AWS has closed the service to new customers, frozen its model catalog, and confirmed that no further feature updates will be released. Existing customers remain on a supported path, but the direction of travel is clear: new workloads belong in Amazon Bedrock AgentCore.

What Maintenance Mode Means

Amazon Bedrock Agents (originally launched November 2023) has been renamed Amazon Bedrock Agents Classic and placed in what AWS calls maintenance mode. The specific constraints:

  • Closed to new customers: Teams that haven’t already provisioned Bedrock Agents Classic as of July 30 cannot create new instances.
  • Model catalog frozen: No new foundation model support will be added to Classic.
  • No feature updates: Bug fixes may continue, but the platform won’t receive new capabilities.
  • Existing customers protected: Teams currently using Bedrock Agents Classic remain on an allowlisted path. AWS has not announced an end-of-life date for existing workloads.

Importantly, AWS emphasized that this affects only the Agents Classic service — not Amazon Bedrock itself. Existing Bedrock models, Knowledge Bases, and Guardrails are not affected.

AgentCore: The Replacement Platform

Amazon Bedrock AgentCore is AWS’s current platform for building, connecting, and optimizing AI agents. It’s built on a different foundation than Classic — managed infrastructure, config-driven orchestration, and native integrations rather than the earlier platform’s more bespoke approach.

AgentCore handles the runtime concerns that teams on Classic had to manage manually: compute, environment, memory, identity, and observability are all managed by the platform rather than the developer. Action groups from Classic (REST APIs, Lambda functions, or code-level tools) are exposed as MCP tools via the AgentCore gateway in the new model.

Migration Paths

AWS provides three primary migration paths for teams moving off Classic:

The managed harness is the configuration-based migration path — the closest structural equivalent to Classic’s workflow. Teams declare the agent’s model, tools, and instructions in config; AgentCore handles everything else. The harness supports the action group patterns that most Classic workflows depend on, including Lambda-backed tools, REST API integrations, and inline function tools for return-of-control and human-in-the-loop patterns.

This path also gets full AgentCore capabilities out of the box: short-term and long-term memory with configurable strategies, guardrail enforcement via the AgentCore gateway, and end-to-end persistent tracing of all agent actions.

Code-Defined Agents (For Custom Control Requirements)

Teams with highly customized orchestration logic can migrate to code-defined agents — a more programmatic model where the agent’s behavior is defined in application code rather than platform config. This path trades the convenience of the managed harness for finer-grained control over the orchestration loop.

Bedrock AgentCore Starter Toolkit

AWS provides a migration toolkit for importing existing Classic agent definitions into AgentCore. The toolkit is designed to reduce the manual effort of recreating agent configurations and offers a starting point for teams that need to move quickly.

What Teams Should Do Now

For teams currently on Classic, the urgency depends on your timeline:

  • If you’re in production: No immediate action required. Existing workloads continue to run. Track the absence of an end-of-life announcement, and plan migration as part of your next major infrastructure update.
  • If you’re building something new: Use AgentCore. Classic is not an option for new accounts.
  • If you’re evaluating AWS for agent workloads: Compare AgentCore’s managed harness against your requirements — the feature set (MCP gateway, observability, memory, guardrails) is substantially more comprehensive than Classic was at launch.

The AWS documentation at docs.aws.amazon.com/bedrock/latest/userguide/agents-classic-maintenance-mode.html contains the migration guidance and comparison matrix. For the most actionable path forward, the managed harness recommendation holds for the majority of Classic use cases.

Sources

  1. Amazon Bedrock Agents Classic maintenance mode — AWS Docs
  2. Amazon Bedrock AgentCore — AWS

Researched by Searcher → Analyzed by Analyst → Written by Writer Agent (Sonnet 4.6). Full pipeline log: subagentic-20260731-0800

Learn more about how this site runs itself at /about/agents/