A team develops a shared summarization prompt in Amazon Bedrock Prompt Management. Production must keep using an approved configuration while developers continue improving the draft.
Identify the prompt configuration used by the production application, and evaluate later changes before applying them.
Which approach best manages the production reference?
Iterate on the draft and explicitly reference a version created from the approved configuration in production. Recording its ARN identifies the applied configuration, but does not guarantee identical generated output or continued model availability.
Detailed explanation
A version is a static snapshot of a configuration, separating the production reference from the working draft.
CreatePromptVersion creates a static snapshot of the draft and returns information including the version ARN. Production can reference the approved snapshot while later draft changes become a separately evaluated version.
This does not separate ongoing development from the approved production configuration.
Using one editable reference for both approved production behavior and ongoing development makes changes harder to control. Developers can continue editing the draft while the application explicitly uses an approved version.
A name alone does not identify the approved configuration snapshot.
A prompt name helps identify the managed resource, but that resource can have multiple versions. Link the approval record and application reference to the specific version representing the approved configuration.
Version numbers identify creation order; they do not guarantee approval or quality.
Keeping frequent snapshots is not inherently a problem. The error is treating creation order as an adoption decision: experimental configurations could reach production without evaluation or approval.
Check official documentation
Compare the following points with the official documentation.
Read AIF-C01 Task 3.2.5, Deploy a prompt, Create a version, and CreatePromptVersion. Sketch the draft, a version, and the application reference as separate elements.
Official references
- Official documentation 1
- Official documentation 2
- Official documentation 3
- Official documentation 4
- Official documentation 5
What to confirm
Explain how production can reference an approved version ARN while draft editing continues, and distinguish version numbering, approval, and identical generated output.
Key points
- Draft versus snapshot
- Explicit version ARN
- Configuration versus output
Notes
- Environment: Reference-based review of current AIF-C01 Task 3.2 and Amazon Bedrock documentation using invented configurations; no AWS accounts, API calls, or model inference.
- Check the static-snapshot description and the returned version ID and ARN in the official documentation; no API call is required.
- Separate creating a version from changing an application reference. Do not infer approval or model availability from the version number.
Foundation review
Create a production version from the draft
In Prompt Management, edit and save the draft, then create a version when the intended configuration is ready. Creating a snapshot preserves that configuration; deciding whether it is approved remains part of the release process.
Make the applied version explicit
Associate the returned version ID and ARN with approval records and the application configuration. Creating a new version does not by itself mean an existing application switches to it.
Distinguish configuration identity from identical output
Runtime variable values and generation variability still matter when using the same prompt version. Identifying the same stored configuration does not guarantee that every response is identical.