A pluggable platform for A/B-testing goal-oriented conversation agents against benchmark scenario packs and a calibrated customer simulator.
BlogWhy you can't grade a conversation with an answer key — and how we evaluate persuasion agents as A/B "matches," framed through Cialdini's Influence. Accents on the dataset, metrics, and methodology.
Field guideThe illustrated, end-to-end field guide: the pluggable engine registry, the customer simulator, domain packs, the dataset, and how the pieces fit together.
Reference docs (Markdown): Architecture · Plugin guide · API