Aisop – Define AI agent workflows as Mermaid or JSON flow graphs (github.com)

The launch of AISOP, an open protocol for defining structured AI programs through Mermaid or JSON flow graphs, marks a significant advancement in the AI/ML community. This innovative framework allows developers to create multi-step AI workflows characterized by various control flows, including branching, parallel execution, and error handling, all structured in a portable JSON format. Notably, AISOP files are machine-readable and can be seamlessly edited in any text editor, allowing for greater interoperability and version control through standard git practices, which enhances collaboration among developers. The dual support for both visual (Mermaid) and structural (JSON) representations enables flexibility, while its token efficiency—using approximately 50% fewer tokens than traditional JSON formats—can significantly improve processing speed and reduce costs associated with AI operations. AISOP's architecture features a three-layer separation of metadata, flow graphs, and function definitions, alongside 14 control flow patterns and robust error handling mechanisms. This makes it an adaptable and comprehensive tool that can work across diverse AI runtimes, crucial for building responsive and responsible AI systems aligned with the principle of human centricity outlined in its foundational Axiom 0.