Open Facilitation Library

AI is starting to run group discussions: it asks the questions, writes up what people said and reports what a group agrees on. There is no shared, open way to tell whether it ran a facilitation method faithfully, or which decisions should stay with the people taking part. Each vendor checks its own tool.

The Open Facilitation Library builds that shared layer in the open:

• Method specs: an open registry of facilitation methods (Delphi, Dynamic Facilitation, Appreciative Inquiry, Sociocracy 3.0 patterns and more), each described stage by stage with the criteria for whether it ran well, written to be portable across platforms.
• Conformance model: a way to check an AI facilitator against a method that separates what the method requires from one product's style, and records who reviewed each criterion.
• Evaluations: open tools for scoring transcripts and whole processes across sessions.
• Facilitator review: practitioners review and change the criteria at openfac.org/calibrate.

OFL grew out of Metagov's AI & Facilitation seminar series. Harmonica is its reference runtime.

openfac.org · wiki.openfac.org · github.com/Open-Facilitation-Library