{"slug": "aws-bedrock-agents-classic-vs-bedrock-agentcore-whats-the-difference", "title": "AWS Bedrock Agents (Classic) vs. Bedrock AgentCore — What’s the Difference?", "summary": "AWS has rebranded its original Bedrock Agents service as 'Classic' and is moving it into maintenance mode, closed to new customers after July 30, 2026, while introducing Bedrock AgentCore, a modular infrastructure platform that lets developers bring their own agent code and control orchestration. AgentCore, launched in preview May 2026 and GA June 2026, offers 12 components including persistent memory, a tool gateway with OAuth, and native observability, whereas Classic provides a fully managed, opinionated service with AWS-controlled orchestration and session-scoped memory. The shift means teams needing custom orchestration, multi-agent systems, or non-Bedrock models should adopt AgentCore, while those with simple, working setups can stay on Classic.", "body_md": "If you’re building on AWS Bedrock, there’s a big shift happening in how agents get built. Here’s the breakdown.\n\n𝗧𝗵𝗲 𝗢𝗻𝗲-𝗟𝗶𝗻𝗲𝗿\n\n→ Bedrock Agents (Classic): a fully managed, opinionated agent service. AWS controls the orchestration loop. → Bedrock AgentCore: a modular infrastructure platform. You bring your own agent code and control the orchestration loop yourself.\n\n𝗕𝗲𝗱𝗿𝗼𝗰𝗸 𝗔𝗴𝗲𝗻𝘁𝘀 (𝗖𝗹𝗮𝘀𝘀𝗶𝗰)\n\nLaunched November 2023, now rebranded “Classic” and moving into maintenance mode — closed to new customers after July 30, 2026.\n\nWhat you get:\n\nAWS-managed ReAct-style orchestration loop\n\nAction Groups (OpenAPI schemas or Lambda functions) for tool calls\n\nBuilt-in RAG via Knowledge Bases\n\nManaged session memory (short-term, within a conversation)\n\nLaunched in preview May 2026, GA June 2026 — this is AWS’s forward direction for agent infrastructure.\n\nYou bring your own agent code — any framework: Strands Agents, LangGraph, CrewAI, LlamaIndex, or a fully custom build — and AgentCore supplies the surrounding infrastructure through 12 components:\n\nRuntime — deploys your agent in serverless, session-isolated containers with auto-scaling\n\nHarness — an optional thin orchestration layer if you want AWS to manage the loop\n\nMemory — persistent, long-term state across sessions (not just session-scoped)\n\nGateway — managed tool connectivity with auth and rate limiting\n\nIdentity — OAuth2/OIDC auth for tools, with token refresh and per-user permissions\n\nCode Interpreter — sandboxed code execution\n\nBrowser — web browsing capability\n\nObservability — real-time tracing, logging, and metrics\n\nPayments — support for agent-initiated transactions\n\nEvaluations — testing and scoring against datasets\n\nPolicy — guardrails and access control\n\nRegistry — a catalog for discovering and reusing agents\n\n𝗞𝗲𝘆 𝗗𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝗰𝗲𝘀 𝗮𝘁 𝗮 𝗚𝗹𝗮𝗻𝗰𝗲\n\nPhilosophy: Classic handles everything for you. AgentCore hands you the infrastructure and lets you own the agent logic.\n\nOrchestration: Classic is fixed and AWS-managed. AgentCore is fully yours to write, in any framework.\n\nModels: Classic is Bedrock-only. AgentCore supports any model, any provider.\n\nMemory: Classic is session-scoped. AgentCore is persistent and cross-session.\n\nTools: Classic uses Action Groups. AgentCore uses a Gateway with managed OAuth and rate limiting.\n\nMulti-agent: Limited in Classic. First-class in AgentCore.\n\nObservability: Classic relies on CloudWatch. AgentCore has a dedicated tracing and metrics component.\n\nEvaluation: Not built into Classic. Native in AgentCore.\n\nLearning curve: Classic — about 30 minutes to a first agent. AgentCore — steeper, since you need to understand each component.\n\nStatus: Classic is entering maintenance mode. AgentCore is under active development and now GA.\n\n𝗪𝗵𝗲𝗻 𝘁𝗼 𝗨𝘀𝗲 𝗪𝗵𝗮𝘁\n\nBedrock Agents Classic fits best when:\n\nThe current setup is already working well\n\nThe flow is simple: prompt → tool calls → response\n\nNo custom orchestration logic is needed\n\nBedrock-only models are sufficient\n\nZero infrastructure management is the priority\n\nAgentCore fits best when:\n\nCustom orchestration or multi-step planning is required\n\nNon-Bedrock models, or a mix of models, are needed\n\nThe system involves multiple agents\n\nPersistent memory across sessions is required\n\nFine-grained, per-user tool authentication is needed\n\nA framework like LangGraph, CrewAI, or Strands is in use\n\nProduction-grade observability and evaluations are required\n\nThe deployment is for enterprise use with VPC/PrivateLink requirements\n\n𝗧𝗵𝗲 𝗧𝗮𝗸𝗲𝗮𝘄𝗮𝘆\n\nBedrock Agents Classic was AWS’s “easy button” for agents. AgentCore is the platform for teams that have outgrown that button and want real control over orchestration, memory, and infrastructure — without giving up managed deployment, observability, and security.", "url": "https://wpnews.pro/news/aws-bedrock-agents-classic-vs-bedrock-agentcore-whats-the-difference", "canonical_source": "https://blog.devgenius.io/aws-bedrock-agents-classic-vs-bedrock-agentcore-whats-the-difference-439387e3b567?source=rss----4e2c1156667e---4", "published_at": "2026-08-09 04:06:01+00:00", "updated_at": "2026-08-09 12:10:14.403804+00:00", "lang": "en", "topics": ["ai-agents", "ai-infrastructure", "ai-products", "ai-tools"], "entities": ["AWS", "Bedrock Agents Classic", "Bedrock AgentCore", "LangGraph", "CrewAI", "LlamaIndex", "Strands Agents"], "alternates": {"html": "https://wpnews.pro/news/aws-bedrock-agents-classic-vs-bedrock-agentcore-whats-the-difference", "markdown": "https://wpnews.pro/news/aws-bedrock-agents-classic-vs-bedrock-agentcore-whats-the-difference.md", "text": "https://wpnews.pro/news/aws-bedrock-agents-classic-vs-bedrock-agentcore-whats-the-difference.txt", "jsonld": "https://wpnews.pro/news/aws-bedrock-agents-classic-vs-bedrock-agentcore-whats-the-difference.jsonld"}}