Machine Learning
Amazon Augmented AI
Amazon Augmented AI (A2I) adds managed human-review workflows to machine learning predictions. It routes uncertain or high-impact GenAI outputs to reviewers and stores auditable review results.
Key points
- Human loops can be started through integrated services or custom applications
- Worker task templates present context and capture structured decisions
- Private workforces keep sensitive review tasks within an organization
When to use it
- Review low-confidence document extractions before RAG ingestion
- Approve model-generated decisions in a regulated workflow
Exam tips
- Choose A2I when a managed human-review loop is required; Step Functions is broader orchestration
- Give reviewers evidence and clear rubrics, because agreement without context is not meaningful oversight