Personal story

My personal journey at KONE HQ

I spent several years working at KONE's headquarters in Espoo, Finland. It's a large, well-run corporate campus, modern, well-maintained, the kind of building where there is an existing way of managing people flows, services, and employee experience.

Many large office buildings run on assumptions. Facility teams know how many people badge in. They have floor plans and meeting room bookings. What nobody has is the real-time picture of how people actually move, how spaces are genuinely being used, and where the building is silently wasting resources, money and employee time every single day.

We started working on a pilot to measure people movement across areas and leverage simulations to address key challenges key areas.

Cost reduction

The most immediate business case is cost

In most commercial buildings, a large share of operating costs is driven by systems running on fixed schedules, not actual usage. Empty floors stay fully serviced. Food is being prepared based on fixed plans. Systems run at peak long after people have left. Facilities keep working for activity that isn't there. Building systems and lighting follow the same clock, conditioning and illuminating zones that cleared out hours earlier. Badge-in data tells you who arrived, not where people actually gathered, which floors emptied first, or when the building was truly done for the day. The waste compounds quietly across energy bills, kitchen spoilage, and contractor hours — with no single view tying spend back to real occupancy.

Shifting to demand-based operations using real-time people flow data can cut these costs by 15–30%.

At campus scale, that's a meaningful P&L lever — not a rounding error.

The multiplier

The real multiplier when occupancy data talks directly to the building

Measuring occupancy is useful. Wiring that data into systems that act on it in real time is where the value multiplies.

The clearest example comes from our work at KONE HQ. Elevators.

KONE HQ uses destination control, already a significant step beyond conventional call buttons. Passengers enter their destination floor before boarding, the system groups them intelligently and assigns the most efficient car. It's smart. But it's still reactive: it responds to inputs as they arrive.

What changes when you add real-time lobby and floor occupancy data is the timing. Destination control can see that demand is coming before passengers have even entered their floor destination. A wave of arrivals detected in the lobby, rush hour building on the ground floor, the system pre-positions cars and distributes load before the queue forms rather than after.

The second gain is skipping unnecessary stops. When a cabin is empty and occupancy data confirms no one is waiting at the floor it's heading to, it doesn't stop. It continues to where demand actually is. A small optimisation that, multiplied across hundreds of journeys per day, meaningfully reduces average travel time.

The result: even a system that is already intelligent becomes measurably faster. The morning bottleneck that everyone accepted as an inevitable feature of working in a tall building turned out to still be a data problem, just a more subtle one.

This is the difference between a Vision Sensor deployment and an operational intelligence platform. Vision Sensors are how the building sees itself. The integrations are how it acts on what it sees.

The operator

Reporting and control without another dashboard

At KONE HQ, the pilot started with measurement: where people move, when floors fill and empty, how lobby demand builds before it hits the lifts. The Building AI Operator is the layer that turns that picture into routine work — the reports facilities teams used to assemble by hand, and the controls that should follow occupancy instead of a fixed clock.

Reporting. Instead of pulling badge counts, BMS exports, and lift logs into a weekly slide deck, the operator drafts what the team actually needs from the same live stream. A morning digest lands in the channel the facilities team already uses: which floors are active, where queues are building, how elevator performance compared to yesterday's peak. Weekly summaries tie equipment events to crowd context — not "lift bank B out of service at 08:12" alone, but what people flow looked like on the floors that bank serves. Monthly utilisation views show which zones stayed busy, which stayed empty, and whether catering and cleaning spend matched real use. The building's history lives in one place, so answering "is this Tuesday normal?" does not require opening five systems.

Control. Reporting alone does not fix the cost problem described above. The operator also acts where integrations allow, always at a safe priority beneath life-safety and fire layers, with every command logged and reversible. At KONE HQ, the clearest control path is vertical transport: pre-positioning cars before queues form, and skipping stops when data shows no one is waiting or elevator car is full. Escalations, maintenance tickets, and tenant-facing issues still route to humans; the operator handles triage, context, and the routine commands the team would have issued if they had been watching every floor at once.

Most of the time the team does not log into a new dashboard. They ask in plain language, approve what matters, and let the operator run the transcription work. That is the shift at campus scale: one teammate who sees the building in real time, reports what happened, and controls what can safely follow occupancy — so the people on site spend their time on judgement, not on stitching data together.

Next steps

Getting started

Our pilot is still running at KONE HQ, and each month of data makes the recommendations sharper.

The Building AI Operator can be deployed in a matter of days or weeks — our sensors capture movement and distance, not images or faces, so there is no DPIA, no works council negotiation, no IT security review for video data. Deployment friction is dramatically lower than camera-based alternatives.

If you manage a large office portfolio and want to understand what your buildings are actually doing, we're always up for a direct conversation.