Amazon Bedrock AgentCore
Infrastructure for running AI agents
A set of services to deploy and operate AI agents at production scale with any framework or model: Runtime for running agents serverlessly, Memory, Gateway to turn APIs into tools, Identity, Browser, Code Interpreter, and Observability.
- Cost$$ Medium
- Difficulty●●●●●
- Costs money when idle?No
- Serverless
Think of it asLike a ready-built industrial park for robots: power, water, security, and storage are all there, you just bring your robots in to work.
When to use it
✓ Use it when
Taking agents from prototype to production with security, context memory, and monitoring.
✗ Avoid it when
Simple model calls or Bedrock's built-in agents are enough → Bedrock.
Common uses
- Customer support agents that call internal APIs
- Agents that browse the web and run analysis code
- Identity management when agents access third-party tools
What does it cost?
- Pricing model: Charged by usage of each component: CPU time and memory while the agent runs, number of memory events, number of tool calls via Gateway…
