Capital in legacy industries isn’t flowing to the obvious AI plays.
Three categories I’m seeing investors drawn to:
1. Teams that deeply understand a specific analog industry and are building AI-native systems of record, not horizontal tools slapped on top of old workflows.
Construction is the clearest example right now. In Q2 alone, 68% of the $3.96 billion flowing into built environment tech went to companies baking AI/ML into their core product. Companies like Buildots are replacing the entire way progress gets tracked on site:
Crews wear 360° cameras, the system compares footage to the schedule automatically, and delays get flagged without anyone filing a report. The product IS the workflow, not a layer on top of it.
2. Startups that can underwrite real-world behavior using unique data exhaust - similar to how earlier fintechs used transaction data - and create new credit, insurance, or revenue-financing primitives around SMBs.
A live version of this in legacy sectors is EquipmentShare: its T3 telematics platform tracks where equipment is, how it’s used, and idle time, and that data is used by partners like Leif Assurance to offer cheaper insurance.
3. Companies that show their product is disinflationary for end customers and wage-enhancing for good operators (aligning) with the broader thesis that AI should materially improve outcomes for the working and middle class.
Historically, software has always made one true at the expense of the other. Either it cuts costs by squeezing labor, or it makes workers' lives easier while inflating the P&L. AI tools in this category materially improve outcomes for the people in working-class industries, not just the equity holders.
You can already see this in fleet and logistics:
Customers of Samsara’s AI-powered safety and routing stack report lower accident rates and operating costs, and some - like Mohawk’s supply-chain group - explicitly say those savings let them raise driver pay and regularly award safety bonuses while staying more profitable.
The common thread across all three:
Durable AI businesses in legacy sectors aren’t selling “tools” but transforming outdated workflows so profits improve, customers pay less, and the best operators earn more.