A new arXiv preprint (2610.14818) proposes a reference architecture for integrating agentic AI into enterprise software. The abstract describes agentic AI as moving generative AI applications toward systems that coordinate reasoning, enterprise context, model inference, and external actions.

The architecture's title highlights three concerns: responsibilities, interactions, and variability. The abstract says integration introduces diverse challenges but does not enumerate them in the announcement. With only this source, there are no differing accounts to compare.

The paper appears aimed at giving enterprise software teams a structured way to think about agentic AI components and their relationships, rather than prescribing a single implementation.