The term „compositional AI“ originates from the fields of Artificial Intelligence, Digital Transformation, and Automation. Compositional AI describes a method where various small, specialised AI models are combined to solve complex tasks. Unlike traditional „monolithic“ AI systems that aim to do everything at once, with compositional AI, many small units share the workload.
Imagine a modular system: for every sub-task there is its own specialised AI module. If these are cleverly put together, the overall system can handle tasks that would be too difficult for a single AI model.
A practical example: In automated customer service, individual AIs are used – one for understanding language, another for analysing sentiment in the text, and a third for selecting suitable responses. The combination of these specialised „building blocks“ results in a flexible and particularly powerful system that can be easily adapted to different requirements.
Compositional AI thus facilitates the development of smarter, more versatile applications, making complex technologies more accessible to businesses of all sizes.
Deutsch 
English (UK)