# System planning
<a name="system-planning"></a>

 The system planning focus area provides best practices for designing the AI system stack introduced in the Dataset Planning focus area. The AI system stack consists of training and auxiliary datasets designed in Dataset Planning, and the AI system solving your use case. The AI system itself decomposes into a *core AI system* and an optional set of *filters* (also referred to as *guardrails*) that block or augment data as it flows through, into, or out of the core AI system. The primary design strategies for designing the AI system stack to meet your release criteria are: 
+  Design choices that directly impact the core AI system (baking) 
+  Designing data filters (filtering) 
+  Communicating proper usage to users through documentation, legal terms of usage, classes, or similar channels (guiding) 

 Refer back to [Dataset planning](dataset-planning.md) for best practices that apply to designing training and auxiliary datasets. The Monitoring focus area addresses guiding. Best practices for ML engineering and for optimizing overall ML performance can be found in the [Machine Learning Lens](https://docs.aws.amazon.com/wellarchitected/latest/machine-learning-lens/machine-learning-lens.html). 

**Topics**
+ [AI system architecture design](raisp01.md)
+ [AI system baking](raisp02.md)
+ [Filtering](raisp03.md)
+ [Choosing a system configuration](raisp04.md)