Seeing Through Walls with WiFi: beQ Visits the VENTURE AI PODCAST

Our VP Sales & Operations, Sebastian Ruppert, spoke with Norman Müller from the Bundesverband für KI-Transformation (Federal Association for AI Transformation) about a question that occupies us every day: How do you make buildings smart without surveilling the people inside them?

Outside, the world has been mapped. Satellites, GPS, and maps deliver real-time data for almost any location on Earth. But the moment you step into a building, that transparency ends at the nearest wall. That's exactly where beQ comes in — and exactly what Norman Müller wanted to discuss in the latest episode of the VENTURE AI PODCAST. His opening question had teeth: Is this a privacy-friendly technological leap, or just the next elegantly packaged form of surveillance?

A fair question. And one we're happy to answer.

From University Experiment to Business Model

Our tagline is: seeing through walls with WiFi. That sounds like a stretch at first — which is exactly why we chose it. But there's solid science behind it. It started with a university experiment showing that off-the-shelf WiFi routers generate signal data on the side that works much like sonar. When a person or object moves through the signal field, the signal changes measurably. Until now, this data simply went unused. Norman put it nicely during the conversation: recycled data waste.

Building on this foundation, we developed a technology. A building's existing WiFi infrastructure is supplemented with our hardware unit, M-QUBE, and the beQ software. That's all it takes — no cameras, no additional sensors, no new cabling. Installation works remotely, and from then on the system delivers real-time, analyzable information on occupancy counts, movement, dwell times, or the position of objects. People appear exclusively as anonymous silhouettes.

The decisive boost came from artificial intelligence. Only AI models turn the faint signal changes into reliable insights, and only they make the solution efficient enough to become broadly affordable. For the end customer, the system costs around 10 cents per square meter per month. For 4,000 square meters of retail space, that's 400 euros. We're convinced that a technology only truly helps people once it's not just something large corporations can afford.

What You Can Do With It

During the conversation we went through several use cases, and every industry has its own. A shopping center can identify which routes visitors take, where they stop, and for how long. In many places, someone still sits at the entrance with a handheld counter to do this. A hotel knows a room was vacated at nine in the morning and can send housekeeping earlier instead of waiting until official checkout. A logistics company can track where goods in its warehouse disappear to, without fitting every single item with a sensor.

And then there are the cases where it's about more than efficiency. A nursing home can detect when a resident has fallen and has been lying on the floor for half an hour. In a fire, rescue teams can see whether and where people are still in the building, instead of having to open all 198 rooms individually to find out.

The Uncomfortable Questions

Norman didn't make it easy for us — and that was a good thing. His example: at a shipping company, Klaus is sitting on the forklift; his shift is known, so his silhouette could, in principle, be identified. Isn't that surveillance after all?

Our answer has two parts. First: the data our system outputs is anonymous by default. Where identification is genuinely needed — in care settings, for example — the system works pseudonymized, and only the operator can establish the link to a person, using their own data, which isn't part of our system. Second: warehouses and workplaces already have camera surveillance today, with everything that entails. We're no worse than what already exists — we're the more privacy-friendly alternative to it. And any employer who wants no analysis at all simply won't get one with our system.

The question of responsibility also came up: what if the software reports the building is empty during a fire, and that's wrong? Our position is clear. beQ provides information but doesn't make decisions. The fire department will continue to check every room, just as it does today. The difference is that they can go first to where our system still shows people. That doesn't make the existing process worse anywhere — it improves it. The final decision stays with a human, in an emergency just as in everyday business. Contracts, in the end, are signed by a managing director — not by artificial intelligence.

Where the Data Stays

One question from the conversation deserves its own paragraph, because we get asked it often: what does beQ actually do with all the data? The answer is short: nothing. The data stays decentralized with the customer, on their own infrastructure. There's no central data pool on our end that could someday become a liability, and we've deliberately chosen not to become a data broker or a consulting firm. Our core is the technology, and that's what we keep developing.

With Home-QUBE, our version for private use, we go a step further. The most common question in presentations is, in essence, whether you could use it to spy on your neighbors. You can't, and that's by design: every user has to upload their own apartment's floor plan, and the system only operates within those boundaries.

What's Next

In July, we released our MVP; testing with initial pilot customers is starting now. Broad market launch is planned for early 2028. Anyone who wants to try the technology early in their own company, or stay informed, can find the waitlist link in the podcast episode's show notes — or reach out to us directly.

The full episode, with every question — including the ones we didn't fit in here — is available here: https://www.youtube.com/watch?v=-E1UG-DoJdI

Our thanks go to Norman Müller and the Bundesverband für KI-Transformation, of which we've been a member for over a year, for a conversation that asked the right questions.