L1 Human Context Layer: dynamic behavioral profiles for each participant. L2 Relationship Graph: verified real-world interaction records. L3 Group Intelligence Engine: N-person group chemistry prediction. L4 Execution Layer: end-to-end meeting coordination logistics. L5 Outcome Feedback Loop: continuous training data collection from real-world outcomes.
The Group Intelligence Engine optimizes for emergent group dynamics, not just pairwise compatibility. It considers energy dynamics (balanced introductive/connective mix), role diversity, goal complementarity, conversational breadth, and cultural compatibility. It is trained on observed group chemistry scores from real Weekend Club events — outcome data that no other system has access to.
Recommendation algorithms are trained on digital behavioral signals (clicks, views, follows) and optimize for engagement. The Relationship Brain is trained on real-world outcome data (group chemistry scores, attendance verification, follow-up behavior) and optimizes for interaction quality. These are fundamentally different training signals producing fundamentally different prediction capabilities.
AI agents integrate through the Relationship Brain API: before coordinating a real-world interaction, the agent submits participant profiles and context to the compatibility endpoint and receives structured compatibility scores, trust flags, group formation recommendations, and no-show risk assessments. After the interaction, agents can submit outcome data to contribute to the shared behavioral graph.
Individual behavioral records are protected under The Weekend Club's privacy framework, which includes differential privacy principles and purpose limitation. API consumers receive aggregated compatibility scores and trust signals, not raw behavioral records. Participant consent is required for profile inclusion, and participants can request data deletion.