Abstract

A technique is proposed herein for automatically generating workflow descriptions in an agentic or multi-agent system. The technique iteratively generates ordered, one-line descriptions of essential workflow steps from an input schema and uses a critique agent and a flexible convergence model to evaluate and refine those descriptions. A multimodal artificial intelligence (AI) agent uses the resulting descriptions and a graph representing workflow execution to generate a final workflow description, while application targets can be determined from schema metadata, reasoned inference, or multi-label classification.

Creative Commons License

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.

Share

COinS