Latest / Future of Work Tech with Fexingo: Remote Tools, AI Productivity, and Workplace Software / How Digital Twins Are Reshaping the Workplace
Transcript
- Lucas: So there's a concept from manufacturing that's quietly taken over the corporate office world, and it's called the digital twin. Basically, it's a virtual replica of a physical space — factory floor, office floor, warehouse — that updates in real time with data from sensors, cameras, and badge swipes. Luna: I've heard of this in aerospace, like NASA using digital twins for spacecraft. But for a regular office? What does that actually look like day to day? Lucas: Right. Let's make it concrete. There's a mid-sized tech firm — about 1,200 employees, hybrid schedule — that built a digital twin of its headquarters last year. They fed in data from desk sensors, Wi-Fi connection logs, and calendar integrations. Then they used the model to test a new seating arrangement and a staggered start-time policy before implementing anything physically. Lucas: The result: they cut desk idle time by 23 percent. That means fewer empty desks, less wasted real estate, and employees reported no drop in collaboration. In fact, the model predicted that certain teams would actually bump into each other more often in the new layout, and that turned out to be true. Luna: Twenty-three percent is not nothing. But I'm wondering — if today was actually useful to you, it's worth mentioning that the way we keep this show ad-free is through listener support. If you'd like to help, buy me a coffee dot com slash fexingo. It's a small thing that makes a big difference. Lucas: Absolutely. And that support lets us dig into stories like this without any sponsor influence. So back to the digital twin — the key is that it's not just a static 3D model. It's a live system. You can run simulations: what happens if we switch to a four-day week? What if we bring back 20 percent more people on Tuesdays? The model shows you congestion patterns, energy use, even noise levels. Luna: Is this something only big companies can afford? I imagine the sensor setup alone is expensive. Lucas: It's getting cheaper. Five years ago, a digital twin project for a mid-size office could run seven figures. Now there are SaaS platforms that start around $50,000 a year. The sensors themselves have come down — a desk occupancy sensor is maybe $15 each if you buy in bulk. And some companies are using existing Wi-Fi data, so they don't need new hardware. Lucas: But there's a bigger question here, and it's the one I find most interesting: what happens when you start modeling not just desks and walls, but people's actual behavior? Because that's where the privacy tension shows up. Luna: Right. If the model knows that Sarah from accounting always goes to the third-floor kitchen at 10:15, is that data anonymous or is it just one step away from identifying her? Lucas: Exactly. And that's the line companies are still figuring out. The firm I mentioned — they anonymized the data at the aggregation level, so the model only sees group patterns, not individual movements. But not every company is that careful. There have been cases where managers used occupancy data to track specific employees, which is a huge red flag. Luna: Do you think there's a risk of the digital twin introducing bias? If the model is trained on historical data from a period when the office was mostly in-person, and now you're using it to design a hybrid schedule, aren't you baking in old assumptions? Lucas: That's a really sharp point. Yes. Let's take a concrete example: if your historical data comes from 2019, when everyone commuted five days a week, the model might optimize for maximum desk occupancy. But in 2026, that's not the goal. You might actually want some desks empty to support collaboration zones. So you have to retrain the model with current behavior, and even then, you have to decide what 'optimal' means. Lucas: And bias can creep in through the sensor placement, too. If you put more sensors in the executive wing than in the open-plan area, the model will have a richer picture of executives' movements and might over-index on their preferences. Luna: So who decides what the model optimizes for? Is it the facilities manager, the HR team, or the CEO? Lucas: Often it's a cross-functional team, but in practice the CFO often has a big say because the business case is about real estate savings. And that's where the tension between cost-cutting and employee experience gets interesting. A digital twin might tell you that you can reduce your footprint by 30 percent, but if you do that without considering how people actually work, you'll end up with a cramped, noisy office that nobody wants to come into. Luna: I've seen that happen. A company I know downsized from three floors to two based on utilization data, but they didn't account for the fact that people from different teams had started overlapping on the same days. It got so crowded that half the team started working from coffee shops. Lucas: Right. And a digital twin could have prevented that — if they'd run the simulation with actual calendar data and team adjacency preferences. That's the promise. The technology is sophisticated enough to model 'what if we put the design team next to the product team?' and see how interaction patterns change. Lucas: Ford has been doing this for years in their factories. They built a digital twin of their Dearborn truck plant, and they use it to simulate assembly line changes before moving a single robot. Amazon does it in their warehouses — they model picker routes to minimize walking time. The office world is just catching up. Luna: So where's the bleeding edge? What's the next step beyond desk sensors? Lucas: Some companies are experimenting with digital twins of entire cities. Singapore has a national digital twin called Virtual Singapore that integrates data from government agencies, traffic sensors, and even weather stations. They use it to plan urban development, emergency response, and energy efficiency. On a smaller scale, there are office parks in Texas that have digital twins of the whole campus — including parking lots, cafeterias, and HVAC systems. Lucas: But the next frontier for the workplace specifically is probably the 'human digital twin' — a model of an individual employee's work patterns, preferences, and even biometrics. Some wellness startups are already offering this: a digital twin that suggests the best time for you to take a break, or which meeting room would be least distracting based on your focus history. Luna: That sounds like it could cross a line fast. If my employer has a model of my attention span, what stops them from using it to schedule my day for maximum output, ignoring my actual preferences? Lucas: Nothing, theoretically. And that's exactly the conversation that needs to happen before this becomes mainstream. The technology itself is neutral — it's a tool for optimization. But the values baked into the optimization are a human choice. Right now, most companies using digital twins are transparent about the data they collect and give employees opt-out options. But as the models get more detailed, the pressure to use them for performance management will grow. Luna: Is there any regulation on the horizon? GDPR covers personal data, but does it cover a digital twin of your behavior? Lucas: That's a gray area. The European Union's AI Act, which came into force earlier this year, classifies some workplace surveillance AI as high-risk, which means it has to undergo conformity assessments. But a digital twin that only models aggregated occupancy data probably falls below that threshold. Individual employee twins, though — if they're used to make decisions about promotions or scheduling — would definitely be high-risk. So the regulatory framework is starting to catch up, but it's patchy. Lucas: I think the real test will come in the next couple of years, when a company tries to use a digital twin to justify layoffs or reorganizations. If the model says 'this department is underutilized by 40 percent,' and the company uses that to cut headcount, you'll see legal challenges about whether the model was accurate and fair. Luna: That feels inevitable. So for a listener who's thinking about implementing a digital twin at their own company, what's the one piece of advice you'd give? Lucas: Start with a narrow question. Don't try to model everything. Pick one pain point — say, 'our desks are empty on Fridays, but we don't know if that's because people are remote or because they're meeting in the cafeteria.' Build a small digital twin of just that area, validate it against real observations, and then expand. That's what the mid-sized tech firm did, and it worked. Lucas: And involve employees from the start. If people feel like the twin is being used to watch them, they'll game the system — they'll badge in and walk away, or they'll disable their Wi-Fi. But if they see it as a tool to make the office better for everyone, you get better data and better outcomes. Luna: So transparency is actually a feature, not a bug. Lucas: Exactly. The companies that do this well treat the digital twin as a shared resource, not a surveillance tool. And that's the difference between a technology that improves work and one that erodes trust.