Sure! Here’s the translation into American English:
—
When implementing artificial intelligence agents in the Amazon Bedrock AgentCore Runtime, currently in preview, customers often look for the ability to use custom domain names. This option allows them to provide a more professional and cohesive experience. By default, the agents of the AgentCore Runtime use endpoints like “https://bedrock-agentcore.Source.amazonaws.com/runtimes/Source/invocations.” However, users can transform these links into more user-friendly custom domains by using Amazon CloudFront as a reverse proxy, and combining Amazon Route 53 and AWS Certificate Manager (ACM) to ensure a secure, scalable custom domain configuration.
Amazon Bedrock AgentCore Runtime is framework-agnostic, allowing integration with various solutions such as LangGraph, CrewAI, or custom-built agents. It’s known for its extended runtime, which can last up to eight hours, making it ideal for tasks requiring complex reasoning. Additionally, each user session operates in its own isolated microVM, providing the necessary security for enterprise applications. Its pricing model is consumption-based, allowing users to only pay for what they use. Unlike other hosting solutions, Bedrock includes integrated authentication and specialized monitoring capabilities for AI agents.
Using a custom domain has multiple benefits. Not only does it enhance the branding experience by exposing a custom domain instead of AWS infrastructure details in network requests, but it also simplifies maintenance, especially when deploying multiple agents or updating configurations. This streamlines the developer experience by making it easier to use the endpoints.
The workflow for implementing this solution begins with a client application that authenticates using Amazon Cognito to receive a token. It then makes an HTTPS request to its custom domain. Route 53 resolves this DNS request to CloudFront, which directs it to the Amazon Bedrock Runtime agent. This way, both the application and backend endpoints can use the same CloudFront, avoiding cross-origin resource sharing (CORS) issues.
To implement this custom solution, several steps are required, ranging from creating a directory and initializing an AWS CDK project to handling RSA and creating the CloudFront distribution that will act as the proxy.
Implementing a custom domain not only enhances the user and development experience but also improves infrastructure management. This enables businesses to focus on creating advanced solutions without worrying about the technical details of hosting, offering additional benefits such as cost reduction and improved access to CORS-related configurations. Thus, an effective and scalable method for managing AI agents in enterprise environments is established.
Referrer: MiMub in Spanish
