The buzz around generative design and architecture would have you believe that in the near future, architects will conceptualize soaring skyscrapers with the ease of plugging in a prompt for a chatbot. But Adam Tank doesn’t think the technology needs to be used for fantastical structures and incredible feats of engineering.

As Chief Customer Officer at Transcend, a New Jersey-based software firm focussed on improving infrastructure design efficiency, Tank says the near-future of AI is better suited to design projects that are, admittedly, much more boring in scope.

Tank and his team are working with AI to make plumbing and water treatment systems more streamlined, with generative AI doing most of the design heavy lifting. In Brazil, Transcend worked with BRK Ambiental, a firm that owns and operates more than 160 wastewater treatment plants across the country, where roughly half of the country still lacks in-home sewage collection. As BRK continues to expand by acquiring and buying new systems, it’s “perpetually in a state of planning,” says Tank. That takes a considerable amount of money and energy the company doesn’t have.

Enter generative design. The solution BRK adopted, the Transcend Design Generator, rapidly prototypes buildings and infrastructure and helps to accelerate approvals and designs. Caique Amorim, a BRK sanitation engineer, says Transcend software allowed BRK engineers to run unlimited scenarios and quickly devise solutions. The savings from using Transcend to do preliminary design have been so significant, BRK stopped hiring engineering consultants to do site planning and design work, and instead brought this work in-house.

By using the system to figure out the most efficient plans to install new water treatment infrastructure, BRK can cut the time it takes to create a design from four months to a week, and also prepare needed documentation for bank loans and environmental approvals. That can mean the difference between a favela, or informal settlement, being able to afford a new treatment plant or not.

Infrastructure’s case for generative AI

Many other firms utilize similar technology to speed up other large-scale infrastructure work. ALICE Technologies helped improve construction efficiency on a massive railway development in Northern England. Montgomery County, Maryland, used a program called Remix to more quickly create street design alternatives. And a Spanish utility used a program called PVcase to more rapidly prototype a new solar power plant.

Generative design is becoming a much bigger part of the rote, but important, parts of architecture and infrastructure design. U.S. construction spending, which hit nearly $2 trillion last year, shot up 7.4%, in large part due to the 17.6% growth in infrastructure spending. With the U.S. also short 133,000 engineers, the advantage in automating repetitive tasks and stretching manpower offer Tank and Transcend a business opportunity.

Transcend said that its generator has been used to design water treatment plants in more than 70 countries. The firm has also designed a module that designs electric substations; Tank hopes that it can play a role in speeding up electrical infrastructure expansion at a time when renewables and manufacturing developments are putting strain on the existing power grid. And it’s also included a software update that can redesign water treatment plants to be PFAS compliant and help remove the dangerous chemicals, a new requirement the EPA just assigned to utilities.

“The critical infrastructure world–rods, bridges, wastewater, power, the not that sexy stuff-there are no that many companies playing in the space today,” says Tank. “I wish there were more.”