Permits Delay Ex-Amazon Execs Building Houses

The triple-decker on Gilman Street in Somerville, Massachusetts went up in four days. Twenty-four factory-finished modules, toilets and all, arrived on the bed of a truck and were stacked by a crane. The robot had framed the walls inside each box at a rate of 30 linear feet per hour — roughly the output of two experienced carpenters — for a total of more than a thousand linear feet before a single module left the Andover microfactory where they were made.

The permit took nearly 10 months.

That gap — between what the machine can do and what the system around it allows — is the real story of Reframe Systems. The Somerville triple-decker, a three-story building with factory-finished volumetric modules, is one of the most compelling demonstrations in years that robots might actually help solve the housing crisis. It is also a useful reminder that housing is not primarily a manufacturing problem.

Reframe was founded in 2022 by three former Amazon Robotics leaders — Vikas Enti, Felipe Polido, and Aaron Small — who collectively oversaw the deployment of more than 500,000 robots across Amazon's fulfillment network. Enti, the CEO, spent years watching robots work at the scale of a single warehouse. Housing, he reasoned, was the same problem with higher stakes. "With a 4.7 million home shortage, the US can't afford incremental gains," he told Robotics and Automation News.

His microfactory model is architecturally distinct from the traditional modular housing factory. A conventional modular framing cell costs 5 million to 10 million dollars and takes up most of a building. Reframe's robotic work cell fits in 500 square feet, costs 200,000 dollars, and can be deployed in under 100 days. The company currently automates roughly 20 percent of framing tasks — walls and ceilings — with plans to reach 60 to 80 percent of all factory tasks eventually, including finishing work like painting, drywall, and window installation. Its Andover microfactory is 16,000 square feet. A second facility, the company says, will cost less than 5 million dollars and is on track to deliver 20 homes this year.

The economics, on paper, are punchy. Developers in comparable infill markets pay 350 to 450 dollars per square foot today. Reframe charges 275 to 325 dollars per square foot — and claims it can deliver a home for about 20 percent less than comparable stick-built construction. The company says it takes 64 minutes of human labor per square foot versus 150 minutes for traditional stick-building. Its long-term target is under 100 dollars per square foot. The robotic cell, Enti told Offsite Builder, "frames at the rate of two carpenters" and meets throughput needs of 30 linear feet per hour.

The triple-decker on Gilman Street — 24 modules assembled in four days after the concrete foundation was poured — cost roughly 1.2 million dollars, or about 300 dollars per square foot, not including land. The customer, according to Slate, got a price that was less than what traditional builders quoted. The company has completed two accessory dwelling units in Arlington and Woburn, and is currently assembling two additional three-unit multifamily buildings in Somerville. A 12-unit development in Devens, Massachusetts — six duplexes, each unit 2,100 square feet with four bedrooms — is expected to deliver its first four homes in July 2026, with the remaining eight between August and October. MassDevelopment, the state economic development authority, is co-developing that project. Reframe won the 2025 Ivory Prize for Housing Affordability in Construction and Design, worth 100,000 dollars, for its work.

But the modular construction industry is not impressed. At the Modular Building Institute's 2025 conference, the CEO of Amherst, a leading modular home manufacturer, made the case directly: "It's pure fiction that you can build cheaper in a factory than on site. All people who work in the factory get benefits, but the competition doesn't have to offer any of that. Modular has a huge cost disadvantage in labor." His company has been building modular homes for decades. Reframe's counter-argument — that robotics and a largely novice workforce (60 percent of its current team are novices or technical high school students in co-op programs; the goal is 90 percent) can close that gap — is plausible in principle and unproven at scale.

The Katerra precedent hangs over the entire sector. The heavily hyped construction startup burned through 2 billion dollars in just six years and went bankrupt, leaving a wreckage of unfinished projects and soured investors. Reframe's 20 million dollar Series A, co-led by Eclipse and VoLo Earth Ventures, is real money but modest by construction-tech standards. The company's stated goal — one million homes worldwide by 2045 — is the kind of number that either looks visionary in hindsight or reckless in the present. Probably both.

There are things the robot cannot do. It cannot speed up the permit process in Somerville, where the building department spent nearly 10 months approving a project that went together in four days. It cannot change the insurance market, which prices factory-built homes differently from site-built ones, sometimes unfavorably. It cannot, by itself, solve the labor shortage — the construction industry is short roughly 500,000 workers in the United States, and a 200,000-dollar robotic cell that replaces two carpenters does not, by itself, train the next generation of framers. Sixty percent of Reframe's workforce are novices or high school students in co-op programs, which is either a feature or a risk, depending on who you ask.

It can, however, catch zoning violations that local officers miss. Enti told Offsite Builder that the company's AI has flagged three such issues in projects to date — a genuine value-add that speaks to one of modular construction's underappreciated advantages: the factory is a closed system where computer vision can see everything, and software can enforce compliance at scale that human inspectors, moving fast across a job site, sometimes cannot.

The 13 Gilman Street building is now occupied. The robots worked. The question is whether the robots working is the hard part, or whether the hard part is everything else — the permits, the insurance underwriters, the labor market, the land, the regulatory machine that took 10 months to approve something that took four days to build. Enti and his team have made a genuine case that factory automation can bend the cost curve on housing. They have not yet made the case that it can bend the rest of it.

What to watch: whether Reframe's Devens project — 12 income-restricted units, measured against traditional stick-built comparables — provides the first real apples-to-apples cost data. The company wants to end 2026 with roughly 60 homes fully built. That number, if it materializes, will tell us more than any pitch deck.