<?xml version="1.0" encoding="UTF-8"?><oembed><type>video</type><version>1.0</version><html>&lt;iframe src=&quot;https://www.loom.com/embed/0632c671ea124d37b76a0b243a36103e&quot; frameborder=&quot;0&quot; width=&quot;1920&quot; height=&quot;1440&quot; webkitallowfullscreen mozallowfullscreen allowfullscreen&gt;&lt;/iframe&gt;</html><height>1440</height><width>1920</width><provider_name>Loom</provider_name><provider_url>https://www.loom.com</provider_url><thumbnail_height>1440</thumbnail_height><thumbnail_width>1920</thumbnail_width><thumbnail_url>https://cdn.loom.com/sessions/thumbnails/0632c671ea124d37b76a0b243a36103e-0d060c06fa7b5f2e.gif</thumbnail_url><duration>292.138667</duration><title>Grid Orchestration with Human Approval Safety</title><description>This Loom explains how a home-battery orchestration system handles ERCOT scarcity events with human-in-the-loop approvals. It simulates 10,000 batteries dispatching over 8,000 hours on a real ERCOT Scarcity evening using real settlement data, committing 36,000 kilowatts and showing gaps like 166 homes and 573 kilowatts dropping out mid-event. The fleet prices the resulting hole at $4,812 at risk and then recovers $4,751 while preserving member backup and using an audit trail for manual overrides. It also demonstrates security and safety checks, rejecting an agent’s attempted overcapacity claim and rolling back a firmware-capability test affecting 100 homes with zero reserve violations. The speaker concludes that outcomes are roughly break-even in held-out tests and notes the code is public on GitHub at Regant/Slash Grid Signal.</description></oembed>