Free tools Windows power users keep installed
One-click scans. No signup required.
Amazon Bedrock AgentCore is AWS-managed infrastructure for deploying and operating AI agents. AWS introduced it in preview on July 16, 2025, as a modular set of services for running agents built with different frameworks and models; AWS announced general availability on October 13, 2025. It is therefore no longer a preview service, although specific features and regional availability can change.
What Amazon Bedrock AgentCore does
AgentCore supplies infrastructure for the operational work around an AI agent: running it, retaining useful context, connecting it to tools, controlling access, and monitoring its execution. AWS positioned it for production concerns such as session isolation, long-running execution, secure browser and code use, and troubleshooting—not as a particular agent model or a requirement to use only Amazon Bedrock models.
AWS described the services as modular and composable. A team can use the components together or select individual services, and can use open-source agent frameworks and models hosted outside Amazon Bedrock. That flexibility does not remove the need to choose a framework, design the agent’s behavior, define authorization, or decide what data it may access.
Which services are included?
The initial service lineup AWS announced in July 2025 included seven components. The descriptions below reflect AWS’s launch materials; capabilities and limits can evolve after launch.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWindows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstall#1 Best Overall
| Service | What it does | How it fits an agent deployment |
|---|---|---|
| Runtime | Managed, low-latency execution with session isolation. AWS’s initial service description said it supported long-running workloads up to eight hours. | Runs agent workloads without requiring teams to manage all the underlying execution infrastructure themselves. Treat the eight-hour figure as a launch-era description, not an assurance that every current workload or configuration has that limit. |
| Memory | Managed short-term and long-term memory across interactions. | Supports retaining context beyond a single interaction, subject to how the application chooses to store and use it. |
| Gateway | Turns existing APIs and services into agent-ready tools and supports MCP-oriented integration. | Provides a way to connect agents to tools and services rather than building every integration as a bespoke component. |
| Browser Tool | A secure cloud browser runtime for web-based services and complex web tasks. | Enables browser-mediated tasks in a managed environment; teams still need to set appropriate access and safeguards for the sites and actions involved. |
| Code Interpreter | A sandbox for executing code across multiple languages. | Lets an agent run code in an isolated environment rather than relying on an unrestricted application host. |
| Observability | Execution visibility, dashboards, metrics, and troubleshooting support, powered through CloudWatch and compatible with OpenTelemetry. | Helps operators inspect agent activity and diagnose issues. CloudWatch and other supporting AWS services may have charges separate from AgentCore usage. |
| Identity | Integrates with identity providers including Amazon Cognito, Microsoft Entra ID, and Okta. | Supports controlled authorization when agents access AWS resources and third-party tools. |
How AgentCore helps move an agent toward production
A working prototype often proves that a model can complete a task, but production deployment also requires predictable execution, access controls, integrations, and operational visibility. AgentCore’s components address those infrastructure needs in separate layers:
- Execution: Runtime provides managed execution and session isolation; Memory addresses context that must persist across interactions.
- Connectivity: Gateway connects APIs and services as tools, while Browser Tool supports web tasks and Code Interpreter provides a sandbox for code execution.
- Control and operations: Identity supports authorization, and Observability provides execution visibility and troubleshooting capabilities.
This division lets a team adopt only the services it needs, rather than treating the whole suite as an all-or-nothing agent framework. It also means production readiness remains an application design task: the team must determine which tools an agent can call, what identities and permissions it receives, how memory is used, and how the resulting system is monitored.
Is AgentCore still in preview?
No. AWS announced the preview on July 16, 2025, then announced general availability on October 13, 2025. The GA announcement added or confirmed enterprise capabilities including VPC, AWS PrivateLink, CloudFormation, and resource tagging across services.
AWS release notes have also described an AgentCore CLI public preview, framework integrations, VPC deployment for Runtime and Memory, an official MCP server, evaluation improvements, and compliance updates. A feature labeled public preview is distinct from the AgentCore service as a whole being in preview. Check AWS’s current documentation and release notes for the status and prerequisites of any particular capability before planning around it.
Recommended Free Tools
Rank #3
Regions and availability
At the July 2025 preview launch, AWS listed four regions: US East (N. Virginia), US West (Oregon), Asia Pacific (Sydney), and Europe (Frankfurt). That is the original preview footprint, not a current global availability list. The cited launch and GA announcements do not establish the present region-by-region availability, so check AWS’s current service documentation for the regions and feature combinations you need.
How AgentCore is priced
AWS offered preview use at no charge through September 16, 2025, and began billing AgentCore usage on September 17, 2025. That free preview period has ended. The cited announcement does not provide current unit rates or a representative deployment total, so no single price can be inferred from it.
Budget for the AgentCore services you use and for related AWS services separately. AWS specifically noted that supporting services such as CloudWatch could incur separate charges for Observability. Check the current AgentCore pricing information and the pricing pages for dependent services before estimating total cost; actual spend depends on usage and configuration.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What to evaluate before choosing AgentCore
AgentCore is most useful to compare as an infrastructure layer, not just as another model or agent-building interface. Assess the following against the requirements of your deployment:
Quick Recap
Best Value
- Framework and model flexibility: Confirm that your framework and model-hosting choices are supported for the workflow you intend to run; AWS says AgentCore can be used with open-source frameworks and models outside Amazon Bedrock.
- Execution and memory: Check isolation, workload duration, scaling expectations, and the memory behavior your application requires.
- Tools and data: Determine whether Gateway, MCP-oriented integrations, browser tasks, or code execution cover the connections your agent needs.
- Security and network posture: Validate identity-provider integration, permissions, VPC and PrivateLink requirements, and resource-tagging needs for your environment.
- Operations: Check that the available visibility, metrics, tracing and evaluation capabilities support your team’s troubleshooting and governance practices.
- Total cost and procurement: Estimate AgentCore usage alongside charges for supporting AWS services. AWS also announced Marketplace listings for AI agents and tools, including selected solutions deployable on AgentCore; assess those as procurement options rather than assuming they are required.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
