Context: The value of demand response is decided in narrow windows, when the grid is under pressure and action must be immediate, with participation rising or falling based on what happens in those moments.

Challenges: Forecasts shift but outreach doesn't adjust, messages go out without urgency or relevance, and follow-up is inconsistent, so by the time an event is active, the opportunity to influence behavior has already narrowed.

Solution: The Demand Response Participation Guidance Agent, an AI agent built using Microsoft Copilot Studio, monitors event participation signals as an event unfolds and intervenes during outreach and reinforcement, flagging under-participation risk while the response window is still open.

Impact: Utilities adopting this approach see engagement happen earlier and reinforcement become timely and focused, with participation rising as a result of better timing, not louder outreach.