UtiliTies reads two neighboring utilities' public construction plans — Dominion Energy South Carolina and Georgia Power — and finds where they could coordinate. A team of AI agents parses the filings, places every project on a map, ranks overlapping pairs by distance and timing, and puts a sourced cost and savings range on each one. Code does the math and models do the reading, so every figure traces back to a filing page or a cited source. Built at ShellHacks 2026 with a team of four, where it won the Sperry Tech challenge.
Reproduces all 6 of Sperry's known overlaps exactly — to 0.01 miles and to the day.
Parsed 252 projects from two real regulatory filings and found 72 coordination opportunities, 29 of them built at the same time.
Agents run in parallel on a custom DAG executor, streaming live progress to a map and pipeline graph over server-sent events.