SITTING IN AN UPPER LEFT OFFICE on the banks of the Danube River, Sylwester Pawluk begins to describe to me the digital journey a modern jet engine takes before the plane even leaves the gate.
A Digital Twins Story
As I gaze out the windows on a rainy day at Graphisoft Park, Pawluk quickly gives me his background story. A career at Google, academic life at Oxford University, and his tenure at GE in Hungary all led him to Graphisoft, where he now leads the global product management team in thinking about data, AI, and digital twins.
“Before a jet engine departs,” he says, “it takes a digital journey, with all the factors involved in the planned flight, like weather, wind speeds, fuel, weight, etc.. When the plane actually takes off, it will feed differences to the pilot. And the pilot then responds to that.”
Sylwester Pawluk, Senior Director of Product Management, Graphisoft. He is Graphisoft’s member contributor to the Nemetschek Group’s AI Hub teams (learn more about that below). (Image: Graphisoft)
This kind of development is relatively new in aviation, but it is the kind of thing that is impacting thoughts about digital twins in buildings. Today’s and tomorrow’s buildings are increasingly electronic. Tomorrow’s buildings are living sensors—living in the sense of the digital journey the GE jet engine experiences before an actual flight. But how does this relate to data and architecture?
With any algorithm you have, whatever quality data you put in, you can easily create an output. If you put in rubbish data, you get rubbish data out.
“We can think about the systems in buildings and even as simple as the aspect of humidity,” remarks Pawluk. “Where is the building? Is it close to a river?” as he gestures outside towards the Danube River, which happens to be at record highs due to torrential rains north of Hungary in Europe. He mentions how connected data, including weather data, can help prepare a building’s systems automatically for such weather events, being intelligent enough to know how to adjust systems and recognizing that occupant loads in the building would change a certain way due to the storm, et cetera.
Intelligence is only possible when a building has a digital twin. The digital twin is electronically connected to the building’s many systems (mostly MEP systems, but security, data infrastructure, automated facade systems, lighting controls, shade and window controls, and such can all be connected). Like the jet engine flight that happens inside the digital twin software, a building’s digital twin software can preemptively prepare a building for the hours ahead of it.
Building Lifecycle, Data and AI
The Nemetschek Group, of which Graphisoft is a part, is focused on serving the AEC/O industries with a portfolio of solutions directed at the entire lifecycle of buildings and physical infrastructure. The digital twin is a critical element in the Group’s grand strategy. So is artificial intelligence (AI).
What this means at Graphisoft now is the company is thinking deeply about how data inside Archicad might need to be restructured. “We want to make decisions on how we develop products based on the right data,” says Pawluk.
This is an image of BIMx running on an iPad in an AEC field situation. Graphisoft’s BIM model and document viewing technology is increasingly vital to its larger common data environment (CDE) goals and data story. So much so that the technology is now central to the Nemetschek Group’s larger One Nemetschek Cloud strategy, which we discuss here.
“With any algorithm you have, whatever quality data you put in, you can easily create an output. If you put in rubbish data, you get rubbish data out.”
Pawluk adds: “If you look at the AEC/O industry and its landscape of software products, it is not unified. First of all, the majority of solutions and the older established platforms look at files [as data], not databases.”
Our discussion returns to the jet engine example. The digital twin software running the simulated flight before the flight combines trusted, high-quality data from multiple data lakes. The algorithms may, in some cases, use generative AI to mass produce the thousands of variance simulations of what may happen during the flight. However, at the end of the process, it will be the variances from the actual flight that are fed to the pilot that combine with human intelligence.
If you look at the AEC/O industry and its landscape of software products, it is not unified. First of all, the majority of solutions and the older established platforms look at files [as data], not databases.
For architects and engineers to have similar systems at their disposal, data inside their software tools must detach itself from the confines of “files and file formats” and live in interoperable and API-based databases. “This switch must change for AI to be meaningfully applied,” adds Pawluk. There is little point in AI algorithms if the data they are operating on is somehow faulted.
I asked Pawluk what “rubbish data” would look like in the context of BIM. He confirmed my hunch that if an architect or other AEC/O user wanted to obtain the exact quantities of all the layers of a wall, that person would need to know the precise quantities of all the layers that make up the walls, how those layers may overlap at wall intersections (because even the small inches add up), and importantly, account for the calculations of all penetrations through those walls. Only then would the data output not be rubbish.
Nemetschek and AI
Pawluk is the AI and data lead for Graphisoft. Its parent company, Nemetschek Group in Germany, announced earlier this year a new AI Hub in Munich, Germany. “The AI Hub is part of a larger overarching organizational shift about AI throughout the Nemetschek Group,” he says. “There is a core team in Germany, but the extended team consists of at least one person from each of the relevant daughter companies in the Group.”
MORE: Nemetschek Creates AI Innovation Hub for Construction Industry
These single AI and data staff members in the daughter teams regularly coordinate efforts with the leaders of the AI Hub team back in Germany. Announced in May of 2024, the AI Hub team is led by Charles Sheridan, Chief AI and Data Officer, and Julian Geiger, Vice President of AI Products and Transformation, Nemetschek Group. They are both formerly from Google, where they held senior roles.
Since earlier this year, there has been a new mandate within the Group to capitalize on the shared IP across the group. The first fruits of that strategy have manifested in the AI Visualizer technology, which is now common and shared between Graphisoft, Allplan, and Vectorworks. And Pawluck assures me there are other exciting common plans also under development, but nothing they can reveal yet.
The Nemetschek Group’s AI Visualizer technology first emerged from Graphisoft but is now shared among the three BIM platforms in the Group. From the most basic of 3D geometry, AI Visualizer lets architects text-prompt their way to stunning conceptual visualizations that are informative, heuristic, and inspirational. (Image: Graphisoft)
Meanwhile, other synergies between daughter companies have been underway for a few years, noticeably between the Group’s BIM authoring platforms and Bluebeam. You can now create Bluebeam Sessions from within Archicad. With the Bluebeam Connector Add-on, the Bluebeam Connection Palette installs new tools that Archicad users can deploy to accelerate the process of preparing for and running a Bluebeam Session.
“We are going to make continuous progress with this type of connectivity,” says Pawluk. “I know from talking to customers that everyone uses Bluebeam. There are mutual benefits. You see more synergies happening between the Nemetschek brands. That is the direction we are continuously moving—with respect to Bluebeam, in particular.” (see: Architosh, “Bluebeam CEO Talks to Architosh about Future,” 26 Nov 2024)
Closing Thoughts
As the Group moves forward with a more shared and dispersed technology strategy rooted in no duplication of efforts, there is a noticeable uptick in both spirit and optimism and in the financial results at the Group level. On the flip side, the Nemetschek Group and its daughter companies need to remain vigilant about AI backlash.
The backlash to AI is coming about inside established big brands where customers are still awaiting fixes and solutions to non-AI problems and opportunities. Some of those things have lingered for a long time. And another issue with AI is that many architects base their fees on an hourly basis. They wonder how AI-based automation may erode their overall billings and instead hope for AI to deliver more value to their clients without erosion to their revenues. AEC professionals will be looking for AI technologies to enhance their organization’s core value to their clients.
The AI Hub is part of a larger overarching organizational shift about AI throughout the Nemetschek Group. There is a core team in Germany, but the extended team consists of at least one person from each of the relevant daughter companies in the Group.
Graphisoft is keenly aware of these challenges. Their experimental AI technologies that utilize computer vision, AI Visualizer, and physical model-making demonstrate a new kind of early design phase workflow that one can imagine some clients taking a keen interest in, blending old analog model-making with the latest digital technology.
As we wrapped up our interview and walked to the large windows overlooking the Danube River along Graphisoft’s HQ, I asked him how close the river got to breaching the sandbags guarding the overall campus. “They got very close to the top,” he says. But this is the reason digital twins, AI, and trusted data lakes matter.
It’s these integrated technologies that deliver the foresight to help us prepare for future weather shock events, guarding existing building assets and helping architects better plan for future ones.
