How can developers or users access Amazon Bedrock (for example, through the AWS Management Console, APIs, or SDKs)?

Developers and users can access Amazon Bedrock through three primary methods: the AWS Management Console, AWS Command Line Interface (CLI) or SDKs for programmatic access, and direct API calls. These options provide flexibility depending on whether the user prefers a graphical interface, script-based automation, or integration into existing applications. Each method aligns with AWS’s standard practices, making it familiar to those already working with AWS services.

The AWS Management Console is the most straightforward way to interact with Amazon Bedrock. After logging into the AWS account, users navigate to the Bedrock service page, where they can manage foundation models, configure workflows, and control access permissions. For example, the console allows users to test models like Anthropic’s Claude or Amazon Titan directly in the browser by inputting text prompts and adjusting parameters such as temperature or response length. The console also provides tools for monitoring usage, setting up model access policies via AWS Identity and Access Management (IAM), and reviewing pricing details. This method is ideal for initial experimentation, troubleshooting, or scenarios where a visual interface simplifies tasks like comparing model outputs.

For programmatic access, developers use AWS SDKs (e.g., Boto3 for Python or the AWS SDK for JavaScript) or the AWS CLI. SDKs integrate Bedrock into applications by leveraging AWS’s pre-built libraries. For instance, using Boto3, a developer can invoke a model by initializing a bedrock-runtime client and calling the invoke_model method with parameters like model ID and input data. The AWS CLI offers similar functionality with commands like aws bedrock-runtime invoke-model. Additionally, Bedrock’s REST API endpoints enable direct HTTP requests for environments where SDKs aren’t feasible. All methods require AWS credentials (access keys or IAM roles) and specify the AWS region where Bedrock is available. For example, invoking Claude v2 in us-east-1 would use the endpoint bedrock-runtime.us-east-1.amazonaws.com. This approach suits automation, CI/CD pipelines, or embedding generative AI features into applications.