Internal policies change every month.
Answers must use the latest policy without retraining the model after each update.
Which architecture should be considered first?
Use retrieval to fetch the current policies and pass them as context; fine-tuning is generally for behavior or format adaptation, not frequent knowledge refresh.
Detailed explanation
Fine-tuning does not automatically ingest future updates.
External documents can be updated and retrieved at inference time.
This is unnecessarily expensive and slow for routine document changes.
Length, permissions, and conflicting documents still require control.
Check official documentation
Compare the following points with the official documentation.
Amazon Bedrock Knowledge Bases公式ドキュメントでデータ取り込み、検索、生成の流れを確認する。
Official references
What to confirm
更新頻度の高い知識にはRAG、振る舞いや形式の適応には微調整という使い分けを説明できる。
Key points
- Knowledge freshness
- RAG
- Fine-tuning
Notes
- Environment: AWS公式AIF-C01試験ガイドとAWS公式ドキュメントの確認
- Compare the definition with the AIF-C01 objective and the service documentation before applying it.
- Keep the concept separate from a particular answer choice or implementation step.
Foundation review
Update mechanism
An index refresh changes available context; fine-tuning changes parameters and needs a new evaluated release.
Evidence and behavior
Retrieval helps supply facts, while tuning changes how the model responds or follows a pattern.