'Premortem': AI Tool Catches Costly Construction Do-Overs Before They Occur

For a large multifamily ground-up project in the Sun Belt, the interior designer specified sleek on-off switches embedded in the kitchen countertops for every unit’s garbage disposal.

However, this detail was overlooked by the electrical engineer on the project, whose own specifications included garbage disposals with on-off switches mounted on walls behind the sinks. 

Both of those conflicting instructions made it into the building’s final planning documents.

The result?

Once the building was nearing completion, a team of electricians proceeded to install wall mounts behind every kitchen sink, as specified, said Paul Zeckser, co-founder and CEO of LightTable.

“When the interior construction team came in after them, with their sleek, beautiful counters, they were left wondering how they were going to get power connected to the in-counter switches that were already purchased,” Zeckser said.

The building’s owners, who preferred the interior designer’s vision, opted to rip out all the finished wiring and rewire everything at a cost of $280K.

“When LightTable did a historical review of the building’s drawings and documents, this issue was flagged,” Zeckser said. “If LightTable had been in place before the building broke ground, it would have caught that conflict.”

LightTable is a startup that uses large language models and vision models to identify errors and omissions during the preconstruction process. It can ingest and cross-check hundreds of architectural and engineering drawings, specification documents and building codes, flagging contradictions, missing information, inconsistencies and risks in a few hours, Zeckser said.

Launched in late 2024, LightTable has already reviewed projects for Mill Creek Residential, Greystar and dozens of other developers and general contractors.

John Tynan, chairman and CEO of TynanGroup, a national real estate development services firm specializing in hospitality and commercial developments, said LightTable can compare data at scale much faster than a peer review could.

“The day after we sent project documents to LightTable, we could push the review comments to the design team,” he said. “It was early enough that they had just started construction documents, so we were able to get meaningful comments right away.” 

'Humanly Impossible'

Today, it’s not uncommon for a large building of even a moderate level of complexity to require more than 2,000 pages of specification documents and hundreds of sheets of drawings. These may be created by a dozen or more consultants of various kinds, some in different locations and working for separate companies.

“No matter how smart these people are, it’s humanly impossible for them all to be perfectly coordinated, especially when following the construction industry’s traditional processes,” Zeckser said.

While AutoCAD and Revit software are used across the industry, they don’t automatically keep documents in sync.

“Another way the industry addresses errors and omissions is through weekly calls between the various teams, with from 10 to 30 people on the calls,” Zeckser said. “But it’s still not nearly enough coordination.”

The standard industry solutions to address these issues, which are identified only once construction work is in progress, are requests for information, change orders and often expensive rework, Zeckser said.

“Approximately 5% of all construction costs are generated with this double work,” he said. “There’s just an assumption that during projects, a lot of stuff gets figured out on the fly. In other industries, they would call it waste, but in construction, it’s a contingency that developers carry.”

But now, with the advent of artificial intelligence, better solutions are available. Zeckser said LightTable allows builders and designers to perform “premortems” on projects to identify potential issues before buildings go vertical.

The tool doesn’t just identify issues — it also ranks risks, Zeckser said. The highest are the ones that are going to cost the most money or take the most time to fix.

“LightTable's AI platform caught mistakes in hours that outside experts spent weeks trying to find in the design documents,” said Shawn Poore, vice president of construction at Mill Creek Residential. “Critically, the AI generates actionable guidance on the most urgent issues to correct before they become costly bottlenecks.”     

LightTable also serves as a workflow management tool that allows people to collaborate asynchronously to identify problems in real time and creates accountability, Zeckser said.

“We provide a platform for everyone to look at the same documents, even if it’s not at the same time or in the same order,” he said. “LightTable flags the issues it finds and serves as a tool to assign these issues to humans to resolve, with deadlines attached. It creates a process rather than just a deliverable.”

Breaking Down Silos

LightTable’s team is split evenly between AI engineers and seasoned architects and engineers who are alumni of major firms like Gensler.

“These are senior leaders with decades of experience on construction sites and advanced degrees,” Zeckser said. “They’ve built large multifamily and hospitality projects, and they’ve seen everything and they understand the nuances. Their job is to train the model.”

Zeckser said the constant back-and-forth learning between the AI engineering and construction teams is a differentiator for LightTable.

“Our industry veterans are constantly giving our AI engineers feedback on the product,” Zeckser said. “As a result, we are able to deliver a great tool and provide great service.”

Tireless Adviser

One thing that LightTable isn’t intended to do is replace human judgment.

“It’s a tool for architects and engineers, to help them do their jobs,” Zeckser said.

But it is a virtual adviser that never gets tired and can read thousands of pages of documents, analyze huge quantities of drawings and instantly bone up on local building codes around the world, in every language.

Earlier this year, Zeckser’s team was piloting LightTable with a prospective client in the Middle East that was building a school that included a prayer room.

“By reading the local building codes, LightTable was able to flag that in the city in which the school was to be built, prayer rooms over a certain size needed more than one egress,” Zeckser said.

That isn’t something a human without deep local expertise could have quickly identified, he said.

But it is a LightTable proof point. The technology's early adopters echo its value proposition.

“By integrating LightTable, we can put more peer review tasks in one unified place,” Poore said. “When all of a project’s team members are on the same page, it gets completed with more speed and accuracy and far fewer mistakes to resolve. It has enabled us to reduce both the construction timelines and costs on our projects.”

This article was produced in collaboration between LightTable and Studio B. Bisnow news staff was not involved in the production of this content.

Studio B is Bisnow’s in-house content and design studio. To learn more about how Studio B can help your team, reach out to studio@bisnow.com.