Graphical-Probabilistic Modeling of Generative Flows in LLM-Native Software Systems
For software engineers building LLM-native systems, this provides a foundational abstraction for reasoning about stochastic, prompt-dependent behavior, though the work is preliminary and lacks empirical validation.
The paper proposes Generation Networks, a graphical probabilistic modeling framework for documenting and analyzing generative flows in LLM-native software systems, aiming to bring design-level rigor to this immature field.
Engineering LLM-native software remains a challenging and immature field. Current practice is largely exploratory, relying on experimentation and heuristic techniques such as prompting and context engineering. These, however, are low-level and lack the principled structure needed to support design-level reasoning or analysis. In contrast, traditional software engineering leverages modularity and abstraction to communicate and analyze system behavior. To bring similar rigor to LLM-native development, we propose methods for documenting generative flows and for stating properties of LLM-based software designs. Such methods must account for the stochastic, prompt-dependent behavior of large language models while remaining expressive enough to capture emergent phenomena. Our initial approach is based on graphical probabilistic models, tailored to capture phenomena characteristic of LLM-native systems. This framework -- what we term Generation Networks -- aims to provide a foundation for principled reasoning about generative interactions and system-level properties in LLM-centric software architectures.