Anonymized data collection - Generative AI Application Builder on AWS

Anonymized data collection

This solution includes an option to send anonymized operational metrics to AWS. We use this data to better understand how customers use this solution and related services and products. When invoked, the following information is collected and sent to AWS:

  • Solution ID - The AWS solution identifier.

  • UUID - A randomly generated unique identifier for each Generative AI Application Builder on AWS deployment.

  • Timestamp - The timestamp when the data was collected.

  • New Amazon Kendra Index Created - Indicates whether a new Amazon Kendra index was created.

  • Amazon Kendra Edition - Specifies the Amazon Kendra edition selected.

  • Usecase Type - Indicates the selected use case type: Text or Agent.

  • Model Provider - The selected model provider, either Bedrock or SageMaker AI.

  • RAG Enabled - Indicates whether Retrieval-Augmented Generation (RAG) functionality is enabled.

  • Type of Knowledge Base - If RAG is enabled, specifies the type of knowledge base selected (Bedrock or Kendra).

  • Streaming - If the use case is Text, indicates whether the output is streamed.

  • Verbose - Indicates whether verbose logging is enabled.

  • VPC Enabled - Indicates whether a VPC is enabled.

  • Create VPC - Specifies whether a VPC was created by the solution or provided by the business user.

  • Use Case Deployment Source - Indicates whether the use case was created from the deployment dashboard or as a standalone use case.

  • Model ID - If Bedrock is the selected model provider, specifies the LLM model ID.

  • Inference Profile ID - If Bedrock is the selected model provider, specifies the Bedrock inference profile ID used when the model is deployed in a different region.

  • Guardrails Enabled - If Bedrock is the selected model provider, indicates whether Bedrock guardrails are enabled.

  • Provisioned Model Enabled - Whether a provisioned model is used.

  • Prompt Parameters - Configuration options related to prompt behavior, such as rephrasing, user editing, disambiguation, and input/prompt length limits.

  • Tracing Enabled - If the Agent use case is selected, indicates whether tracing is enabled for the Bedrock Agent.

  • Client-Owned User Pool - Indicates whether the Cognito user pool is client-owned.

  • Feedback Enabled - Indicates whether the user has opted to provide feedback (positive or negative) on LLM responses.

  • Usage Counts - Various metric counts collected from the solution’s custom CloudWatch dashboard, which provides application usage analytics. Examples include LLM input and output token counts.

    Note: Customer prompts and Model Arn are explicitly excluded and not collected.

AWS owns the data gathered through this survey. Data collection is subject to the AWS Privacy Policy. To opt out of this feature, complete the following steps before launching the AWS CloudFormation template.

  1. Download the generative-ai-application-builder-on-aws.template AWS CloudFormation template to your local hard drive.

  2. Open the AWS CloudFormation template with a text editor.

  3. Modify the AWS CloudFormation template mapping section from:

    Mappings:
  Solution:
    Data:
      SendAnonymousUsageData: 'Yes'

    to:

    Mappings:
  Solution:
    Data:
      SendAnonymousUsageData: 'No'

  4. Sign in to the AWS CloudFormation console.

  5. Choose Create stack.

  6. On the Create stack page, specify template section, select Upload a template file.

  7. Under Upload a template file, choose Choose file and select the edited template from your local drive.

  8. Choose Next and follow the steps in Launch the stack in the Deploy the solution section of this guide.

Note

By default, users can opt out of sending anonymous usage data for deployment dashboards only. However, use cases within the solution will continue to send anonymous metrics unless explicitly disabled.

To disable anonymous usage data for all use cases:

  1. Visit the GitHub repository and download the source code for this solution.

  2. In the file use-case-stack.ts, set the value of SendAnonymousUsageData to No.

  3. Follow the instructions in the README.md to deploy your customized solution.