Machine Learning
Amazon Bedrock Prompt Management
Amazon Bedrock Prompt Management stores reusable prompt templates with variables, model configuration, and versions. It separates governed prompt lifecycle from application deployments and supports traceable experimentation.
Key points
- Prompt variants enable controlled comparison during development
- Created versions are immutable release references
- Applications must record the version or ARN used for each response
When to use it
- Manage a regulated extraction prompt across environments
- Share a parameterized support-assistant template among applications
Exam tips
- Choose Prompt Management for reusable prompt assets and AppConfig for general runtime flags
- Version the evaluation set with the prompt and validate every substituted variable at the trust boundary