Skip to main content
Cloud / AWS / Products / Amazon Bedrock AgentCore - AI Agent Runtime

Amazon Bedrock AgentCore - AI Agent Runtime

Amazon Bedrock AgentCore: serverless runtime and services to securely run, scale, govern and observe production AI agents.

Machine Learning
Pricing Model Pay-per-use (consumption-based, per-second)
Availability Multiple regions incl. EU (Frankfurt, Ireland, London, Paris, Stockholm)
Data Sovereignty EU regions available
Reliability N/A SLA

What is Amazon Bedrock AgentCore?

Amazon Bedrock AgentCore is an agentic platform from AWS that lets organizations deploy, scale, govern and observe AI agents securely in production. Rather than a single service, AgentCore offers a composable set of modular services that you can use individually or in combination. The central building block is a serverless runtime that executes agents and tools without infrastructure management, providing complete session isolation and execution windows of up to eight hours.

AgentCore closes the gap between prototype and production operation of AI agents. While agents are quick to build with frameworks, running them in production requires secure hosting, identity and access management, persistent memory, connectivity to enterprise systems, and traceability. AgentCore is framework- and model-agnostic: it works with open-source frameworks such as LangGraph, CrewAI, LlamaIndex and Strands Agents, with any foundation model (including outside Amazon Bedrock), and with common protocols such as the Model Context Protocol (MCP) and Agent-to-Agent (A2A).

Core Features

  • Runtime: Serverless, isolated execution environment for agents and tools with fast cold starts, session isolation and execution windows of up to eight hours for both synchronous and asynchronous workloads.
  • Gateway and Identity: Gateway turns existing APIs, Lambda functions and services into MCP-compatible tools and connects to existing MCP servers. Identity manages authentication and access and is compatible with identity providers such as Amazon Cognito, Okta, Microsoft Entra ID and Auth0.
  • Memory: Short-term memory for multi-turn conversations and long-term memory that persists across sessions and can be shared between agents.
  • Browser, Code Interpreter and Observability: Browser provides an isolated cloud browser environment for web interaction, Code Interpreter a sandbox for code execution. Observability delivers tracing and monitoring based on the OpenTelemetry (OTEL) standard.

Common Use Cases

Running AI agents in production: Organizations move agents built in a framework into a serverless, scalable environment with session isolation, persistent memory and built-in observability, without operating servers themselves.

Tools and MCP servers from existing systems: Through the Gateway, existing APIs, databases and Lambda functions are exposed as MCP tools with minimal effort, so MCP-compatible agents can access existing backends without rewriting code.

Internal agent platform: IT teams provide developers with a governed path to build agents using approved tools, shared memory stores and controlled access to enterprise services, with centralized observability, authentication and compliance.

Benefits

  • Framework-, model- and protocol-agnostic: no lock-in to a single vendor or model.
  • Consumption-based, per-second billing with no upfront costs, and I/O wait time is not charged.
  • EU regions including Frankfurt with full feature support, plus support for VPC, AWS PrivateLink and CloudFormation.

Integration with innFactory

As an AWS Reseller, innFactory supports you with the adoption and operation of this service.

Typical Use Cases

Run and scale production AI agents serverless
Expose existing APIs and Lambda functions as agent tools via MCP
Manage long- and short-term memory for context-aware agents
Govern, trace and monitor agents in production

Frequently Asked Questions

What is Amazon Bedrock AgentCore?

Amazon Bedrock AgentCore is an agentic platform from AWS for running AI agents securely in production. It consists of modular services such as Runtime, Gateway, Memory, Identity, Browser, Code Interpreter and Observability that you can use individually or together. The platform is framework- and model-agnostic and removes the need to manage your own infrastructure.

When should I use Amazon Bedrock AgentCore?

AgentCore is a good fit when you want to take an agent from prototype to scalable production without managing servers yourself. Typical scenarios include serverless hosting of agents with session isolation, exposing existing APIs as tools via the Model Context Protocol (MCP), persistent memory across sessions, and tracing and monitoring in operation. The Runtime supports execution windows of up to eight hours, which also covers asynchronous and long-running agents.

How much does Amazon Bedrock AgentCore cost?

AgentCore is billed on a consumption basis with no upfront costs or minimum fees. The Runtime is billed per second based on the vCPU and memory resources used (1-second minimum), and I/O wait time is not charged. Each service is billed separately by usage. For binding prices, see the official AWS pricing page.

Is Amazon Bedrock AgentCore available in the EU?

Yes. AgentCore is available in several EU regions, including Europe (Frankfurt) with full feature support as well as Ireland, London, Paris and Stockholm. Individual regions may have gaps for newer features. At GA, VPC, AWS PrivateLink, CloudFormation and resource tagging were added across all services, supporting data residency and private network connectivity.

AWS Cloud Expertise

innFactory is an AWS Reseller with certified cloud architects. We provide consulting, implementation, and managed services for AWS.

Similar Products from Other Clouds

Other cloud providers offer comparable services in this category. As a multi-cloud partner, we help you choose the right solution.

Google Cloud

Agent Development Kit (ADK) - Multi-Agent Framework

Agent Development Kit (ADK): Google's open-source framework to build, evaluate, and deploy single- and multi-agent …

Pricing Free / open source (Apache 2.0); …
SLA N/A (framework); SLA depends on the chosen deployment target
Compare →
Azure

Foundry IQ - Knowledge Layer

Foundry IQ is the knowledge and retrieval layer for Microsoft Foundry: agents query unified knowledge bases through a …

Pricing Pay-as-you-go (free tier + free token …
SLA N/A (some features in preview, without SLA)
Compare →
Azure

Foundry Local - On-Device Local AI Runtime

Foundry Local: cross-platform AI runtime that runs models on-device via ONNX Runtime. OpenAI-compatible API, no cloud, …

Pricing Free, no per-token cost
SLA N/A (local execution, no service SLA)
Compare →
Google Cloud

Gemini on Vertex AI: Foundation Models via API

Gemini on Vertex AI: Google's foundation models (Gemini 3, 2.5 Pro, Flash) with long context and multimodality via one …

Pricing Pay-per-use (tokens), plus batch, …
SLA 99.9%
Compare →
Google Cloud

GKE Inference Gateway - LLM Routing on Kubernetes

GKE Inference Gateway: Kubernetes-native gateway for serving generative AI on GKE with LLM-aware routing and …

Pricing No separate product charge, you pay for …
SLA N/A (no dedicated SLA, the GKE SLA applies to the cluster)
Compare →
Azure

Microsoft Discovery - Agentic AI for R&D

Microsoft Discovery is an agentic AI platform on Azure that combines research agents, knowledge graphs, and HPC to …

Pricing Pay-per-use (Azure consumption + User …
SLA N/A
Compare →

68 comparable products found across other clouds.

Ready to start with Amazon Bedrock AgentCore - AI Agent Runtime?

Our certified AWS experts help you with architecture, integration, and optimization.

Schedule Consultation