Adam Sand's Foundation Project Labs is taking an AI-native roofing roll-up public, giving roofers equity instead of private equity exits.
Sand predicts robots on roofs within three to five years, with 30 roofers running 15 crews.
Sand says private equity's three-to-ten-year profit flip won't fund the ten-year payoff his AI-native model needs.
Roofers Should Own the AI Roll-Up
Roofing is a fragmented, labor-heavy industry that private equity is already consolidating, buying companies at three times profit and selling the combined platform at ten. Adam Sand wants roofers to own that consolidation rather than sell out, so Foundation Project Labs supplies tools and operating frameworks while owning neither the software nor the roofing companies. The bet is that an AI-native structure can centralise data and train robots to do the hard work, which he expects on roofs in three to five years. Roofers who join get equity in the public entity, and their data improves the robots for everyone.
Takeaways
Owners are the growth bottleneck. Many roofing companies stall at $2m to $3m because owners keep doing every job, skip their own commission and discount roofs, which Sand calls fictitious savings. A sale changes only the owner's shares, so the business must run without them.
Value is a discount on future earnings. A business earning $1m for 30 predictable years can fetch $10m, or ten years of profit. Buyers pay three years of profit when they expect ten years of earnings, and up to ten years when they expect thirty. Investors buy growth, not margin.
Private equity arbitrages fragmented roofing. Private equity buys ten roofing companies at three times profit, combines them and sells at ten times profit. Fragmentation attracts capital that consolidates tens of thousands of operators into thousands and then hundreds.
Foundation Project Labs wants roofers to own roll-up. Its model is a three-legged stool. It supplies tools and operating frameworks, and it trains roofers, building on HubSpot, Zuper, CompanyCam and JustCall. It owns neither the software nor the roofing companies, and Sand wants roofers to hold equity in the public entity.
Roofing data is valuable and contested. AccuLynx sold to insurer Verisk for $2.34bn, 17 times revenue. Sand objects to software terms that hand roofers' data to insurers and manufacturers, and says Delta runs its own ticket system rather than cede that layer.
AI-native roll-ups centralise intelligence, not hierarchy. Private equity will not fund a ten-year payoff on a five-year fund cycle. An AI-native company feeds a central RAG database, and each participant improves the model for all, as Tesla's cars collectively train full self-driving.
Robots will do roofing's hardest work. Sand expects robots on roofs within ten years, personally in three to five. They would handle 90% of the hard work while roofers do craftsman work and robot setup. Heavy equipment replaced 30 shovel workers with two better-paid operators.
Collecting robot training data
Tap a name to go there · tap a line to see why they are linked
On this episode 2
Show 1
Put cameras on workers in developing countries who have no stake in the outcome.
Pay $20 an hour for housework filmed by a head-mounted phone.
Give contributors equity in the training set so they share in its value.
Treat robots as platforms loaded with training sets, like consoles with cartridges.