DX AI

DX AI powers a wide range of features of DX such as the following:

Data processing

DX uses Claude models through AWS Bedrock. Under DX’s agreements with AWS, customer data is not used for AWS’s training purposes and data retention is disabled. For more information, see AWS Data Protection.

DX retains AI-generated content—including data summaries, AI-generated SQL queries, Agent Experience evaluations, session summaries, and work-type classifications—for the duration of your service contract with DX, or per the terms of your services agreement.

DX applies feature-specific scrubbing and data controls before submitting customer data to an LLM provider. For AI Code Insights, the daemon replaces recognized secret and personal-data patterns before sending session messages to DX. Scrubbing does not guarantee that every sensitive value is detected, so residual personal data or code can reach the provider. See AI Code Insights transcript handling.

Data processed by LLM providers can include:

  • Pull request titles
  • Pull request descriptions
  • Pull request labels
  • Issue titles
  • Issue descriptions
  • Issue labels
  • Issue custom fields
  • Survey comments
  • Playbook strategies
  • Scrubbed coding-agent session messages when Transcripts & Agent Experience is enabled

Disabling DX AI

To disable DX AI, please contact your DX account manager.