Roadmap: How to adopt the Agent System Design practice
This roadmap describes how to adopt the Agent System Design practice as well as common pitfalls.
Relationships
Parent Practices
Main Description

Getting Started

Agent System Design is about the system as a whole and the context in which it will be executed. In a sense, the practice is about the most high-level design decisions. Therefore, it is very important that the stakeholders are aware of the constraints, assumptions, and possibilities within the environment and that these things are sufficiently represented in the design.


Common Pitfalls

The design of the overall system and of its environment has great influence on the capabilities and robustness of the developed solution.

Missing information about execution environment

Making informed design decisions requires a good overview of the execution environment. It is possible, however, that the environment can not be fully known at design time. For instance, the regulatory framework and legal standards, can be in flux and it is not entirely clear what the legal situation will be at the time the system is deployed. In such cases, the designers should develop contingency plans along with the Product Owner and the prospective users of the system. These plans should include possibilities for re-scoping or for inclusion of adaptive features that make the design and implemented solution more robust to these kinds of changes.

Unclear interfaces

External systems can include legacy systems or other systems with poorly-documented or poorly-understood interfaces. Ideally, it is possible to acquire a test system against which the system can be tested and that allows the developers to reverse engineer the interfaces. If that is not possible either, other solutions have to be found to ensure that compatibility with the external component is ensured during testing and deployment. If such risks exist, the should be documented in the Risk List and be communicated to the stakeholders early and openly.