GridShadow

Project Report: GridShadow

Track Chosen

Smart Grid
GridShadow aligns with the Smart Grid track because it focuses on grid reliability, real-time monitoring, and operational decision support for utilities and microgrids during extreme weather.

Problem Statement Addressed

Extreme weather is increasing the frequency and severity of power outages. Utilities and microgrid operators receive weather forecasts and alerts, but they still struggle to translate that information into actionable operational decisions, such as:

Which zones/feeders/assets are most likely to fail in the next 6 to 48 hours
Where to stage crews and resources before impact
What to restore first when outages occur
How to prioritize critical infrastructure (hospitals, shelters, EOCs, water systems) without delaying overall restoration or creating operational confusion

Today, many of these decisions depend on static runbooks, spreadsheets, and tribal knowledge rather than real-time, personalized decision support.

Ideation and Development Process

We started by thinking about what a real operator needs during a storm, not what looks impressive on a dashboard. Our earliest idea was simply “prioritize hospitals first,” but we realized that this only becomes useful when embedded inside the workflow utilities actually follow.

We iterated toward a product shaped by four principles:

Start from live reality: Use real storm signals (forecast + alerts) as the trigger for action, not as a standalone visualization.
Make tradeoffs explicit: Operators must balance restoring the most customers quickly with restoring critical facilities first, so we designed the system around controllable objectives.
Personalization is the moat: Every utility and microgrid has different playbooks, thresholds, constraints, and vocabulary. We made memory the core layer so the same storm produces different plans for different operators.
Operator-grade outputs: Instead of only showing charts, the system must generate action plans and briefs that can be used in a control room or emergency operations center.

We scoped the build to demonstrate an end-to-end operational loop suitable for a hackathon: forecast → risk → plan → brief → learn.

Final Solution and Intended Impact

GridShadow is an AI storm operations copilot for utilities and microgrids. It turns live weather into a ranked risk view and generates a personalized storm preparation and restoration plan grounded in an operator’s runbooks, constraints, and priorities.

Core capabilities

Live storm awareness: Ingests forecast and alerts for a territory and highlights time windows of highest operational risk.
Risk ranking: Scores zones/feeders/assets by likely impact and explains drivers (wind, gusts, precipitation, flooding exposure, vulnerability).
Critical infrastructure-first planning: Generates restoration and load-shedding recommendations that prioritize hospitals and other critical services, with an adjustable tradeoff between “maximize customers restored” and “critical facilities first.”
Supermemory personalization: Learns from uploaded documents (operating playbooks, priority rules, crew resources, incident logs) to tailor plans to each operator’s real procedures and thresholds.
Exportable briefs: Produces a control room ops brief, crew dispatch task list, and a public safety brief for coordination with EOCs and critical facilities.
Learning loop: After the event, logs outcomes and updates vulnerability and procedural notes for future storms.

Intended impact GridShadow helps operators act earlier and more consistently under pressure:

Faster protection and restoration of hospitals and critical infrastructure
Reduced operational confusion during high-stress events
Decisions aligned with established runbooks instead of ad hoc judgment
Continuous improvement across storms instead of repeating the same failure patterns

In short, GridShadow upgrades storm response from reactive and generic to proactive, personalized, and critical-first.

MLH track: Gemini API and Auth0