When GenAI becomes a Team Member

Generative AI (GenAI) is increasingly entering team-based innovation work, yet making an AI tool available does not specify how it should complement human facilitation. This study asks: What design principles enable GenAI to complement human facilitation in team-based innovation workshops? We adopt a longitudinal Design Science Research approach across two multi-firm waves of the Platform Thinking HUB. Wave 1 introduced a freely available process-aware co-thinker; Wave 2 redesigned the intervention through staged conversation scripts, phase-specific AI roles, AI-triggered pauses for team deliberation, interaction guidance, and reusable checkpoints. An inductive analysis of chat traces, surveys, interviews, observations, and workshop artifacts shows that GenAI becomes useful through a designed collaboration process rather than through a stable role or autonomous participation. We derive five design principles for structuring the collaboration journey, protecting human deliberation, assigning phase-specific AI roles, preparing participants while using AI to scaffold interaction, and preserving context through validated checkpoints. The study conceptualizes GenAI as an embedded local facilitator governed by a human meta-facilitator and offers actionable guidance for innovation managers designing AI-enabled workshops.