The CONCORD framework uses real-time speaker verification to capture only the owner's speech. It resolves the resulting information gaps by querying other assistants through a relationship-aware disclosure protocol. This asynchronous A2A communication recovers missing context without recording non-consenting speakers. Practitioners can use this to deploy proactive voice agents in shared social spaces.