Generative AI
Building and using generative AI on AWS, at three levels of abstraction.
| Service | Level | Choose when | Key distinction |
|---|---|---|---|
| Amazon Q | Ready-to-use assistant | Chat/coding help with no build | Fully managed AI assistant |
| Amazon Bedrock | Foundation-model API | Build gen-AI apps on hosted FMs | One API to many models; fine-tune, RAG |
| Amazon SageMaker JumpStart | Model hub in Amazon SageMaker AI | Deploy/fine-tune models yourself | You own the model and infrastructure |
- Abstraction ladder: use an assistant (Q) → call models via API (Bedrock) → run models yourself (JumpStart).