03 July 2026 | Interaction | By Editor Robotics Business NEWS <editor@rbnpress.com>
In this exclusive interview with Robotics Business News, Aurimas Sabulis, Founder and CEO of Dextall, shares how the company is redefining modern construction through robotics, standardized manufacturing, and AI-driven automation. He discusses scaling robotic welding, balancing customization with industrial efficiency, and why stable processes—not technology alone—are the key to successful automation in the construction industry.
Dextall emphasizes that the robotic welding platform is already operating at commercial scale rather than as a pilot project. What were the key factors that allowed you to move from experimentation to real production deployment?
We did not move from experimentation to production. We moved from production to robotics, in that order, and the order is the whole point.
For most of our operating history we welded these connections manually. We ran the line long enough to understand every weld, every tolerance, every failure mode, and every variation that crept in because a human was making a judgment call. Only once the process was boring and predictable did we introduce a robot. That sounds slow. It is slow. It is also the only sequence that actually works.
The other factor is utilization. A robotic cell only pays for itself if it runs. A pilot that welds a handful of components a week to prove a concept does not justify the capital or the engineering effort behind it. We deployed when we had committed backlog that kept the cell loaded, which meant the ROI math was real instead of theoretical. The robot was the last decision in a long chain, not the first.
Your team standardized five structural hook configurations into a single over-engineered design before introducing robotics. Why was standardization more important than automation itself in achieving manufacturing scale?
This is the most important question in our story and I appreciate you asking it.
We had five hook configurations. Each one was defensible on its own. Each one was engineered for a slightly different load case or panel condition. From a structural standpoint it was clean. From a manufacturing standpoint it was a disaster. Five variants meant five fixtures, five weld programs, five inspection protocols, five SKUs in inventory, and constant cognitive load on the floor about which one was being made today.
We made a deliberate decision to over engineer a single hook that satisfied the worst case load condition across all five. That hook is heavier than it needs to be for most applications. We pay a small premium in steel. In exchange we eliminated four fixtures, four weld programs, four inspection procedures, and four lines of inventory. The economy of scale on that single decision was roughly five times. None of that came from a robot. It came from the decision to make one part instead of five.
The robot only made sense after the standardization. If we had bought the robot first and pointed it at five variants, we would have ended up with an expensive way to manufacture defects faster. Automation amplifies whatever process you point it at. If the process is variable, you get variability at higher speed. If the process is stable, you get scale. The boring work is the standardization. The robot is the reward for doing the boring work.
The company claims the robotic welding system delivers three times the speed of manual welding with identical weld quality every time. How transformative is consistency, not just speed, for large scale facade construction?
Speed is the headline. Consistency is the business model.
Three times the speed changes our throughput and our cost per panel. That matters. But speed alone does not change what we can sell or who will buy from us. Consistency does.
In facade work the cost of a defective weld is not the weld. It is the panel that has to be reworked, the truck that misses its delivery window, the labor time at Manhattan high rise project that gets burned waiting for a replacement, and the trust with a general contractor that erodes when a delivery is wrong. Every one of those costs is an order of magnitude larger than the weld itself.
When weld quality is identical every time, the rest of the system reorganizes around that fact. Our inspection regime gets lighter because we are sampling a known stable process instead of verifying a variable one. Our warranty exposure narrows because failure modes converge to a small predictable set. Our installers on site move faster because they trust the panels will fit. Our engineers can certify performance with tighter tolerances because they are not designing around the worst weld they might receive. On a building with ten thousand structural connections holding the envelope to the structure, identical weld quality is not a feature. It is what makes the building safe to design the way we design it.
Construction has historically been one of the hardest industries to automate because every building project is different. How do you balance customization in architecture with the repeatability robotics requires?
Most people frame this wrong. They assume the trade off is between architectural freedom and manufacturing efficiency, and that you have to give up one to get the other. That is not the trade off.
The real trade off is between variation at the component level and variation at the assembly level. You can have wild architectural expression, different geometries, different window patterns, different cladding, different rhythms, as long as the underlying components are standardized. The cladding of a panel can be unique. The hook that attaches panel to structure, does not have to be.
Think of it the way Lego works. The bricks are completely standardized. The buildings you make with them are not. The standardization of the brick is what enables the variety of the building, because the connection logic is solved and the designer is free to focus on composition.
Our system works the same way. The structural connections, the panel frame logic, the attachment points, the tolerances, all of that is standardized and robot friendly. The facade visible part, the materials, the geometry, the openings, the finishes, remains a design variable. Architects do not lose freedom. They lose only the parts of the design problem that should not have been their problem to begin with. The industry mistake for decades has been treating every connection every back-end detail as a custom engineering problem. That is what makes construction unproductive.
Dextall's backlog now exceeds $210 million with projects involving firms like Turner Construction and SOM. Are major developers and contractors becoming more comfortable relying on robotic manufacturing infrastructure for critical building components?
Yes, and the shift is happening faster than I expected even two years ago.
Several forces are converging at once. Labor on high rise facade work in major US markets is harder to staff every year and the skilled trades that remain are increasingly expensive. Schedule risk on conventional curtain wall is no longer acceptable to capital partners who underwrite projects on tight delivery assumptions. Insurance and warranty markets are quietly tightening around inconsistent fabrication. General contractors are being measured by their developer clients on certainty of outcome, not lowest bid.
In that environment, a manufacturer with a controlled factory, repeatable QC data, and reliable delivery becomes a procurement advantage instead of a procurement risk. The conversation we have with Turner, Gilbane, Suffolk and other large developers / GCs we work with is not about whether the robot welds well. It is about whether we can hit the schedule, hold the tolerances, and absorb design coordination without surprise change orders. The robot is one of the answers, but the question is operational.
What has changed in the last two to three years is that the developers and GCs who used to ask "is this proven" are now asking "can you take more of our scope." That is the inflection point. They are not asking us to convince them. They are asking us to scale.
The company repeatedly states that "automation is not a strategy, it's a reward for stability." How does that philosophy differ from the way many industries approach AI and robotics adoption today?
This phrase is the operating philosophy of the company and it is deliberately at odds with how most industries are approaching AI and robotics right now.
The dominant narrative in 2026 is that you adopt AI or robotics first and the productivity follows. I think that gets it exactly backwards. Productivity comes from process clarity. AI and robotics amplify whatever process you point them at. If you point them at a chaotic process, you get chaos at higher speed. If you point them at a stable process, you get scale.
The hard work, the unglamorous, ego bruising, organizationally painful work, is the stabilization. It means standardizing parts that engineers want to keep variable. It means writing down decisions that used to live in someone's head. It means accepting that a slightly over engineered standard part beats a perfectly optimized custom part every time, because the standard part is the one a machine can build. None of this is exciting. None of it makes a good headline. All of it is the actual prerequisite.
Once that work is done, automation is almost trivial. The robot is the easy part. The five year argument inside the company about whether to standardize the hook is the hard part. Most industries are skipping that step right now. They are buying AI tools and robotic cells and expecting transformation. What they get instead is faster execution of unstable processes, which is usually worse than slower execution of unstable processes, because the failures compound at machine speed.
Labor shortages, rising material costs, and sustainability regulations like New York City's Local Law 97 are accelerating pressure on construction firms. Do you believe industrialized prefab construction is becoming inevitable for high rise development?
For high rise development in dense urban markets, yes. The pressures you named are not cyclical. They are structural.
Local Law 97 is the clearest example. The first compliance period is already in active enforcement and the second compliance period beginning in 2030 will impose materially tighter emissions caps that most existing buildings will struggle to meet without significant envelope intervention. New construction does not get a pass. The facade is the single largest driver of operational emissions performance in a high rise and it is also the hardest system to retrofit later. Building it wrong the first time is a thirty year liability.
A controlled factory environment can deliver tighter thermal performance, more consistent air sealing, and better embodied carbon documentation than site built wall. Not because the factory is more virtuous, but because the process is measurable. When you can measure it, you can certify it. When you can certify it, you can finance it. The capital markets are starting to price this in.
Combine that with the labor reality, the senior tradespeople who built New York's last generation of facades are retiring faster than they are being replaced, and the math is straightforward. Industrialized prefab is not a preference. It is the only way the volume of high rise development that NYC, Boston, DC, and other dense markets need is going to get built on schedule, on budget, and in compliance. I am cautious about the word inevitable because nothing in construction is truly inevitable on a short timeline, but on a ten to fifteen year horizon for high rise envelope work in regulated markets the direction seems to be heading one way.
Looking ahead, which parts of the construction workflow do you believe are most likely to be automated next: fabrication, assembly, inspection, logistics, or on site installation, and where do humans remain irreplaceable?
Fabrication is the easy answer because it is the part of the workflow most amenable to controlled environments. Welding, panel assembly, glazing, and finishing in a factory setting will continue to automate quickly, and we are already a long way down that path.
Inspection is the next obvious frontier. Computer vision and sensor fusion can already detect weld defects, dimensional drift, and surface flaws more reliably than human QC at production speed. That technology is mature. I expect rapid adaption progress over the next three to five years.
Logistics is further along than people realize. Optimized loading, sequencing for just in time site delivery, and route planning are largely solved problems borrowed from other industries. The friction is interfacing with the rest of the construction supply chain, which is still analog in many places.
On site assembly is the hardest. The site is unstructured, weather exposed, congested, and constantly changing. Robotics that work in a factory do not survive a Manhattan construction site. We will see assist technologies, exoskeletons, automated lifts, robotic positioning aids, before we see autonomous assembly at scale. My honest view is that on site installation remains a human led activity for the foreseeable future, with robotics in a supporting role.
Where humans remain irreplaceable: design judgment, client and architect relationships, code interpretation, complex problem solving in unstructured environments, and the kind of trade craft that handles the unexpected condition you only find when you open up a wall. Those are not automation problems. Those are the parts of construction that are about people, and they will remain so.
The frame I would offer your readers is that automation is not displacing construction workers. It is moving the work to where it can be done better, which means moving repetitive fabrication into factories and freeing skilled trades to do the high judgment work on site that machines genuinely cannot do. That is a better industry for everyone in it.
