CivilGrid turns the three-day hunt for whose pipe it is into a subscription

Josh Mackanic's founding story is a phone-call story. An unrecorded pipe turned up in one of his excavations at PG&E, and establishing who owned it and whether the crew could tap it took three days and $60,000 in delay ^[2]. The search, not the shovel, is the unit of work CivilGrid is priced against.

Which makes the choice of outcome number worth a second look. The savings PG&E identified are paving costs, spread across 1,600 planned gas distribution projects ^[8], which works out to roughly $37,500 a project ^[9]. That is money saved at the planning desk, not in the trench. The account does not explain the mechanism by which the map produces a paving saving, and the study was run by the same customer that supplied the endorsement quote, from PG&E's Christine Cowsert ^[7]. No customer count, renewal figure, or measured change in strike rates appears anywhere in it ^[14]. Treat it as a live utility doing arithmetic on its own pipeline: more than most Series A decks carry, less than an independent result.

Here is what teams tell themselves they are buying: strike prevention, because 200,000 strikes a year is the harm in the pitch ^[4]. Here is what somebody will actually open it for on Monday: deciding where the pipe goes before the design freezes. Mackanic describes his core user as the person at the very early stages of deciding what gets built ^[11]. Field safety and pre-design engineering are different jobs, and in most utilities and engineering firms they are different budget lines.

The moat claim deserves close reading, because it is not a software claim. Mackanic says better subsurface visibility is not a new idea, and that what nobody had done was collect and secure the data and build relationships with the parties most likely to pay for it ^[12]. Coverage assembled that way is local. A dataset that is complete inside PG&E's territory tells a county three states away nothing about whether its own water district ever handed over a file. Note the ratio: the savings a single utility says it found are about 2.3 times the whole round ^[10]. If that holds anywhere else, the hard part is not finding buyers, it is getting the next record holder to open its drawer, which is roughly Mackanic's own argument for why utilities running lean would rather this came from outside ^[13].

Two axes settle whether this is a purchase or a demo. First, where your delay starts: in the pre-design question of what is down there, or in the field once the crew is already standing on it. Second, whether the record holders in your footprint, the water district and the county and the neighbouring gas utility, are actually in the dataset for the jurisdictions you build in. Both yes, and the subscription retires a recurring scramble that currently gets paid for in delay days. Either one no, and you are buying a very good map of somebody else's ground, then making the same phone calls.