Signature Homes founder and Chairman Dwight Sandlin’s line — “nimbleness without sacrificing time and resources” — may sound like a homebuilder’s pipedream.
That’s, unless you take the time to sit with what’s underneath it.
In this market, “nimble” isn’t an organizational personality trait. It’s not hype; nor is it an abstraction. It’s an operational must-have.
Would-be homebuyers aren’t saying no because they don’t want a home. They’re saying not yet because the math feels fragile and the risk feels personal. A builder who can’t adapt product, options, positioning, and cost quickly enough isn’t just slower. They’re less credible.
Sandlin describes a company trying to remain credible in a hesitant market while adjusting to reality without tearing itself apart. His point is that the old way—redrawing, reissuing, rechecking, and relearning the hard way—burns precisely what builders can least afford right now: time and people.
That’s where the conversation shifts from a single tool or workflow to homebuilding’s chronic blind spot.
Research and development starved
For decades, residential construction has been one of America’s largest physical production engines — and one of its least research-driven. Homebuilding organizations have learned to invest in land, sales, and operations to drive immediate volume.
They have not learned to invest consistently in a formal “learn faster than reality changes” capability — the kind of R&D discipline other industrial sectors treat as non-negotiable. The consequences are plain in the industry’s most stubborn structural gap: the wide disconnect between back-office systems and frontline reality — between IT (Information Technology) and OT (Operational Technology).
The simplest way to describe that disconnect is this: in many builders’ organizations, the systems that “decide” and the systems that “do” are still not wired together. The result is predictable. Decisions are made upstream with incomplete or stale information. The field absorbs the consequences downstream, where change is expensive, time is compressed, and failure is public. The accounting system eventually records the damage, but it doesn’t prevent it.
In an exclusive interview, Higharc co-founder and VP of Special Projects Michael Bergin offered insight into why Sandlin’s “nimbleness” point is so loaded — and why the next phase of homebuilding performance won’t be a marketing or incentives story. It will be a systems story.
“The trajectory of what we’re doing with Higharc Labs is meant to enable more stakeholders in the process of building new homes to have their perspectives or needs resolved in the product model,” Bergin said. “I think [that] is a big challenge across the board for builders.”
That sentence is the thesis. The bottleneck isn’t that builders lack smart people. It’s that the organization’s intelligence arrives in fragments — and those fragments move slowly, informally, and inconsistently. The jobsite receives the final version of the truth, often long after the “truth” should have been corrected.
IT doesn’t talk to OT
Bergin’s example is the kind of thing that makes operators grimace because it’s so familiar.
“Purchasing has an input,” he said. “This window isn’t available, so we’ll need to switch to this window size.” In the traditional model, he explained, “they might draft an email and send it to the drafting team. The drafting team will address it when they do, and that could be months. In the meantime, there is a risk of mistakes happening.”
That’s IT and OT failing to speak the same language. Purchasing knows the supply chain reality. Drafting controls the plan’s reality. Construction executes the physical reality. If the bridge between those realities is an email, the field will pay for it.
Bergin didn’t stop there. He laid out the broader pattern:
“The same happens for inputs that come from sales, and the same happens for inputs that come from code changes or regulatory changes, and the same happens for executive inputs, right?”
Most builders can read that and immediately picture the cascade: a sales-requested tweak, a spec availability change, a code update, a municipality’s interpretation shifting, a leadership decision on options or pricing — and the slow-motion scramble to align every downstream artifact.
This is where “nimbleness” becomes more than speed. It becomes a form of risk management. Bergin framed the opportunity as transforming change management into a disciplined, governed process rather than a human relay race.
“Being able to take all of those perspectives or needs… and translate them into changes in the model that then just have to be approved by someone with the appropriate administrative capability, takes this change management problem and brings us enormous opportunity.”
He was precise about what that opportunity actually is: a builder finally getting to “a single source of truth.” Not a rhetorical one — a functional one.
“So we can be really certain that we have a single source of truth, not just for the design and drafting team, but for everyone across the business,” he said.
And then he added the line that should make every operator lean in:
“Really changes the dynamics of how long it takes to respond to the various changes happening across the business.”
That’s what the IT–OT disconnect looks like from 30,000 feet: the time it takes for reality to become shared truth — and then to become action.
Что дальше?
Now layer in what’s different about 2026.
Bergin is seeing builders become comfortable with AI in a shallow way first, and then hungry for it in a deeper way.
“I’m starting to see a trend across the board among builders,” he said. “They’re using ChatGPT, maybe for just the typical daily work they would do, and they’re getting assistance that feels fairly magical from a computer. And then they’re looking more broadly and saying,, okay, how can I take the same type of capability and apply it to the core of my business?”
That shift matters because homebuilding has historically treated “technology adoption” as either a procurement decision or a cultural preference — early adopters versus laggards. Bergin’s view is more pragmatic: the market is forcing builders to reduce friction and errors faster than people alone can.
He offered a demo that illustrates what’s changing.
“First, we’re taking that image and passing it to one of our AI models,” he said, describing a workflow that translates a floor plan image into a building information model. “And then, 45 seconds later or so, a building information model appears here, representing what would be many hours of drafting work in some contexts, and in others, a full day of work ormore.”
The temptation is to treat that as a “wow” moment. The bigger point is what comes next: how you govern the system so “wow” doesn’t turn into “oops.”
In homebuilding, you don’t get credit for being fast if you’re fast in the wrong direction. Bergin addressed the obvious concern—hallucination and conflation—without handwaving.
“Yes, certainly there are,” he said. His answer wasn’t to pretend the risk doesn’t exist. It was to design the workflow around it. “What’s happening there is that the prompt is being interpreted and then converted into a more structured analysis of the building to make sure that whatever the prompt… is actually viable to execute on,” he explained. And if the request isn’t viable, “it will say, oh, sorry… we can’t do that.”
That’s a quiet but important distinction. In the consumer AI world, the user is entertained or assisted. In the operational world, the system must be comfortable saying no — because the cost of a wrong yes is a schedule slip, a rework bill, an inspection failure, or a warranty claim.
Domain training
Bergin took it further by drawing a line between general-purpose AI and a production-grade system. He described experimenting with a general tool and acknowledged it was “quite amazing” — something “fundamentally impossible a year ago.”
But he emphasized the risk:
“There’s some risk in depending on these systems, where they might present a response that seems confidently right but is actually quite wrong in subtle ways.” He called out the core weakness: “These systems are still struggling in a very extreme way with geometry.”
That word — geometry — is a reminder that homebuilding isn’t a text problem. It’s physical math under constraints.
It also explains why residential construction chronically underinvests in R&D. In industries that spend heavily on R&D, there is a deep respect for deterministic systems, validation, repeatability, and failure modes. Homebuilding, by contrast, has leaned on craft knowledge, experience, and heroic supervision. Those are valuable. They are not scalable R&D disciplines.
Bergin described Higharc’s early development in those terms.
“There were many years… our first three years of development… [based] on creating a system that was deterministic in its inputs, so it wasn’t capable of hallucinating,” he said. “You can always be sure that if you put the same inputs in, you’re always going to get the same outputs on the other side, which is not the case with large language models.”
That’s the key to the “AI plus human discipline” idea. AI can accelerate translation and suggest changes. Human discipline must govern truth, approvals, and standards. Otherwise, you’ve just invented a faster way to drift.
Where interoperability begins
Bergin’s 2026 view is ultimately about bringing more of the organization into that disciplined loop.
“Number one for me in 2026, as far as what’s exciting and what’s going to become available,” is enabling those cross-functional inputs — purchasing, sales, regulatory, executive — to be resolved in the model, not in the hallway.«
His “second” 2026 emphasis is telling, too: townhomes and a push into land complexity.
“We’re working on townhomes,” he said. “The focus on townhomes is also pulling us deeper into supporting land development… because larger buildings on lots that might not be completely flat can be more challenging.”
The implication is that the old separation — land team over here, product team over there, permitting handled as a separate headache — isn’t going to hold. As density rises and entitlement friction grows, builders will need to model risk “further up the pipeline.”
That includes permitting, where the industry’s OT reality has been stuck for years: manual validation, backlog, inconsistent interpretation, and slow cycles. Bergin called it “the permitting and plan validation aspect of the work,” and said it’s “the really rich opportunity” because “essentially all these are geometric or mathematical rule evaluations that need to be done, but they’re done in a very manual way in the traditional process.”
He pointed to precedent:
“Model-based permitting is a well-established practice in the UK,” and “Singapore requires models and does automated permit analysis on a regular basis.”
The point isn’t that U.S. municipalities will suddenly modernize. The point is that, eventually, the permitting bottleneck becomes too expensive for the system to tolerate — and the builders who can provide better model-based assurance will move faster with less drama.
A fully-integrated tech stack
Then comes the piece often left out of “AI in homebuilding” conversations: the existing ecosystem. Bergin put it plainly:
“ERP integrations and connecting back to the existing ecosystem are another big area of focus for us.” He drew a clear boundary on scope — “Once it goes past construction, that’s not really in our scope. We’re focused on pre-construction optimization” — but he also acknowledged the real-world requirement: “To ensure our customers can benefit from the expanded capability set without significant downstream rework, we must integrate with downstream ERP systems.”
That is the IT–OT bridge in one sentence. A model can be brilliant. If it can’t connect to the systems that run purchasing, scheduling, costing, and accounting, the organization fractures again — and the field reverts to improvising.
So what’s the call to action hidden in all of this?
It’s not “use AI.” It’s not “buy software.” It’s a challenge to builders, developers, and capital partners to acknowledge something uncomfortable: homebuilding has tried to scale a complex production system with a thin R&D muscle. The result is a structural lag between decision and execution — and the cost of that lag is now too high.
Dwight Sandlin wants “nimbleness without sacrificing time and resources” because his market is punishing waste. Michael Bergin is describing what it takes to make that real: more stakeholders, fewer handoffs, deterministic truth, governed change, and integration into the operational backbone. That’s not tech worship. That’s a blueprint for closing the IT–OT gap — by treating the home as a product system that can learn.
In 2026, the builders who separate themselves won’t be the ones with the flashiest AI demos. They’ll be the ones with the discipline to turn AI into a controlled advantage: faster cycles, fewer mistakes, quicker response to market signals, and a product that feels current to buyers who are still deciding whether to step off the sidelines.
That’s the moment homebuilding finally starts acting like an industry that invests in how it learns — not just what it builds.
English