Vikas Enti of Reframe Systems: 5 Questions – Commercial Observer

In August, Massachusetts-based Reframe Systems, a physical artificial intelligence housing company, raised $40 million in financing for a major expansion of its home manufacturing business.

Energy Impact Partners led the round, as Reframe aims to apply robotics, software and regional microfactories — which are exactly what they sound like — to an industrialized homebuilding model it said is capable of delivering homes up to three times faster and at more than 35 percent lower cost than traditional construction.

SEE ALSO: Podcast: The Risks CRE Investors Aren’t Pricing In, with Moody’s Jeffrey Havsy

Co-founder and CEO Vikas Enti leads Reframe. He’s a mechanical engineer and former Amazon Robotics executive who spent 11 years helping scale more than 500,000 robots across the e-commerce giant’s fulfillment network.

Last week, Commercial Observer spoke with Enti about his company’s AI- and robotics-driven microfactories and how they can help build a variety of quality homes globally.

This interview has been edited for length and clarity.

Commercial Observer: Reframe has been operating since 2022, but why the recent $40 million funding?

Vikas Enti: We’re a physical AI company where our whole mission is to solve for housing. How do we make housing abundant and resilient? We’ve developed a technology and a business strategy for building housing in microfactories and building a network of these microfactories around the world that produce homes that are all electric, low-embodied carbon and net zero energy by default. 

The recent fundraise is part of our growth activities. We have a healthy backlog of customers we’re building toward and that requires us to expand capacity — to move from our prototype-scale microfactory here in Andover, Mass., called Fab Zero, to Fab One, which is our production-scale microfactory in Billerica, Mass. This funding allows us to continue building the manufacturing capacity that’s required, building out the teams and the capabilities, so we can continue delivering more housing out here in New England and beyond.

To give you a sense of scale that we’re having to get toward, we’ve built about 10 housing units today over the last two and a half years. We’ll have to build about 130 over the next 12 months, and that number triples as we start looking over the next 24 months. 

So needing the capacity was the driver for launching Fab One, but also raising the funding that’s required.

Is Reframe essentially a more AI- and robotics-infused modular home company, or something else? What in your manufacturing process makes your homes better and less expensive?

We are a turnkey, design-build contractor for our customers. We’re the architect, the manufacturer, and we are the general contractor to install different scales of homes — everything from small cabins to five-story apartment buildings. We deliver these in wood volumetric modules. We’re following regular building codes and building standards, and the materials are familiar. 

What is different are the means and methods we use for delivering these homes. The idea of factory-built housing is well established. What we’ve done differently is designed a unique factory that costs 10 times less than a traditional automated factory for housing production, and is significantly smaller. So we can actually place these factories closer to where customers are. 

Our goal is to have our factories within an hour of most metropolitan areas. And the way we’ve gotten to that is the makeup of the founding team. I spent about 11 years at Amazon Robotics, and so did my co-founders. We got to scale Amazon’s fulfillment network to build, bring same-day shipping, and deploy almost half a million robots out there. We’re bringing a lot of lessons from Amazon into housing production.

The challenge we’ve seen with factory-built housing today is that everyone’s treated this as a mass production problem, and they’ve said we have to build a mega-factory or a gigafactory and try to build one standard type of product. So they end up building Accessory Dwelling Units (ADUs) alone, or they try to go after multifamily work. In either case, they end up with a very expensive factory that’s hard to keep utilized. 

Instead, we said, “Let’s design a manufacturing system that allows us to produce any type of home that can go in any lot over time.” 

That’s our technology edge as we design this home system. All this allows us to simultaneously produce ADUs that are going to California and single-family homes that are going to Massachusetts. The same factory can produce ADUs and apartment buildings simultaneously.

Who are your customers?

Our preferred customer is a developer, someone who has a pipeline of infill work. Our ideal customer is someone who’s building homes in existing neighborhoods. They’re building everything from small triple-deckers to 50- to 100-unit apartment buildings. 

Our whole goal is to solve the housing problem, not just to produce more homes in farmland. The goal is to produce housing in existing neighborhoods where people actually want to live, and the way we do that is to get more density.

What is the role of AI as compared to the human workers in these microfactories?

We see robotics and AI as tools in our tool chest here. Our whole goal is to make our people significantly more efficient. Can we have the same person build five times or 10 times more homes in a year? 

We do this across the entire spectrum, not just in the factory, but from the very get-go. From when we do initial feasibility studies for our developer customers, we have software we’ve developed that allows us to quickly analyze the site, go through zoning rules and building codes to understand what can be built and get to that concept design quickly. 

Then we have our own internal AI that we call pixels to parts, which allows us to quickly convert these concept designs into manufacturing-ready designs. We are allowing a small group of engineers to model these buildings significantly faster.

Then it’s to the factory, where we do two things: A software layer with the AI that’s actually providing rich visual instructions to our builders, to the point that high school co-ops and apprentice carpenters have actually been able to do plumbing work and wiring work following instructions on iPads and all the pre-cut wires and pipes that we have.

Then we also have a robotic work cell. Our robotic work cell can frame walls and ceilings using regular two by fours and two by sixes you can get from Home Depot. We solved some really hard challenges there. How do you get a highly precise robot to be able to work with very natural materials? Our end state roadmap is that we see a world where about 65 to 80 percent of the tasks in the factory are automated with the robotic system.

It seems your microfactory home building is an outlier compared to most companies.

I’d say we’re the only company in the world today that is leveraging robotics and AI to produce volumetric homes or volumetric modules in microfactories and to also build homes that are significantly high performance. 

There are other players that are doing pieces of it. We’ve seen few companies that are doing just the robotics aspect to frame walls and open wall panels. There are a couple of companies trying to just go after the idea of microfactories that they can deploy, but they produce analyzed components and not full volumetric modular homes. 

We find that we’ve picked the intersection of applying technology with the microfactory architecture to go after volumetric modular housing, and that what we’re building is the right approach. The whole calculus is how many labor hours you can actually ship on a truck to a job site, not how much material you can ship.

Philip Russo can be reached at prusso@commercialobserver.com.