Abstract
Automating complex business processes with large language models (LLMs) can present challenges, as monolithic prompts containing extensive domain-specific information and instructions can be difficult to develop and evaluate, and LLMs may lack certain adaptive business logic capabilities. The described method can address such challenges by transforming standard business process flowcharts (for example, decision trees or business process diagrams), into an orchestrated workflow. In this approach, a node of the flowchart can be converted into a distinct step for execution by a dedicated artificial intelligence (AI) sub-agent. The sub-agent may receive a targeted prompt containing relevant context from the flowchart's descriptions and can produce a discrete output, for example a YES/NO response, that may be used to guide the workflow to a subsequent step. This technique can facilitate the automation of intricate business logic by decomposing a complex process into a sequence of manageable tasks, leveraging existing process documentation to provide structured, contextual guidance to the AI at each decision point.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Kuligin, Leonid and Tschochohei, Max, "A Method for Converting Business Process Models into Orchestrated AI Agent Workflows", Technical Disclosure Commons, (September 22, 2025)
https://www.tdcommons.org/dpubs_series/8607