The past few years have seen a drastic rise in the capabilities and usage of AI models in the workforce. A common form of AI being used is Large Language Models and yet, not much is known about how exactly people work with these models. Drawing on empirical observation of multiple design sessions, this research established a preliminary set of architypes to characterize the micro-interact patterns of humans and AI collaboration. Specifically, it found that there are five common types of these interactions—Branch, C, Parallel Lines, Triangle, and Circle—with different user’s varying their usage of them. The types of patterns are stable across users despite significant variation in the way they are combined with the relative emphasis of each pattern. This may be related to how much the users are comfortable relying on the AI output (before independent variation) which will be explored in future work. Future work will expand the study done here to include explicit variation in user experience level with a topic and how their trust in the AI effects their usage.