Physical AI for buildings
Physical AI that autonomously operates HVAC on real equipment. Proven in production.
If a car can drive itself, a building can run itself.
Every other system stops at a recommendation. Ours runs the equipment — perceiving, deciding and acting in a closed loop, inside bounds it can never cross.
One control loop · 01 / 04
Reads the live state of the building from the BMS it already has — over BACnet or Modbus. Temperatures, pressures, flows, occupancy, weather. No new sensors.
Supply-air setpoint
24.4°
Held by schedule
Supply fan
1,315 rpm
2.00 kW · from 2.0
Proven in production
Of commercial space running itself. Not a pilot. Not a dashboard.
12 months · every command bounded
Comfort or safety incidents. No safety bound has ever been exceeded.
Candor TechSpace · The Westin
Air-handling units running themselves. Live, across offices and hospitality.
Live across portfolios including
Five levels of building control
Analytics recommend; an operator still turns the knob hours later. EnergyOS writes the setpoint itself, in seconds, and re-decides when conditions change.
Live
Each flash is a setpoint written to real equipment.
Energy vs. baseline today: −19.3% — an outcome of autonomy, not the goal.
Screenshots from the live system at Candor TechSpace, Sector 21, Gurugram — 33 AHUs across two towers, taken 14 October 2026. Not a mock-up.
The whole building at a glance
Every cooling unit on every floor, in one table. The top strip answers the big questions: how many units are running, how many have a problem, how many need a look.
In this snapshot: 33 of 33 units running, zero faults, 9 flagged for a closer look.
Most systems hand you a number. This one writes what is wrong, what could be causing it, and lets the team raise a work order in one click.
Flagged before anyone complained
In the comfort band, last 7 days
How fast a floor cools — learned, not guessed
From the Candor TechSpace deployment, week of 14 October 2026.
Return air 3.0 °C above its 24 °C band with the chilled-water valve at 100% — the unit is giving all it has. Chilled water is arriving warmer than design and leaving only 6 °C warmer.
Could be
Autonomy is earned, in stages
Six weeks read-only. Six weeks supervised. Replayed against your building's own history before it runs live. Hard bounds on every point, verified with your team. Then, and only then, autonomy.
In 12+ months across 68 AHUs, no safety bound has ever been exceeded.
Physical limits enforced by the equipment itself — cannot be overridden by any software, including ours.
Hard limits on every controllable parameter. The system is architecturally prevented from commanding outside them.
Every command is checked against the building's live measured state before it is written.
Full suspension of all system-issued commands, available to your facility team at all times.
Revert the whole building to manual BMS control at any time, without contacting us.
Any operator instruction takes precedence over any system command. Always. No exceptions.
If the connection drops, the building keeps operating safely within its bounds.
Not every building begins with autonomy. Smart Logger is a wireless, battery-powered sensor and one screen — and on its own it is already earning its keep in a hotel, a pharma clean room and a co-working floor.
Hotel · Westin
Hotels heat water with piped natural gas around the clock, to a fixed temperature, whether or not guests are drawing it. Nobody sees the gap between what the boiler delivers and what the rooms need, because nobody is measuring it.
A probe clipped onto the supply line reads the temperature every few minutes and streams it to one screen. The trace makes the waste visible — hours where the loop runs hotter than it has to, and the overnight stretches where it needn't run at all — so the engineering team can trim the setpoint and the firing schedule with confidence, and watch the gas bill follow.
Hotels: find the gas before you cut the setpointHot water supply
Water temperature at 00:36
Fig. 1 — Water temperature, last 1 h
Wireless, battery-powered, placed or clipped on. No BMS, no wiring, no shutdown.
Water temperature, room temperature, humidity, CO₂ and energy meters — ten sheets, one register.
Loggers establish the baseline. Once the waste is visible and measured, EnergyOS can take over the loop.
Live at The Westin Gurgaon, a pharmaceutical clean room and a co-working operator in the NCR. Savings are reported per site after a full billing cycle.
No. SKW EnergyOS sits on top of the controls already installed — it connects to the controllers inside each air-handling unit over BACnet or Modbus. Nothing is ripped out, and AMC contracts are untouched.
No. The goal is the right temperature at the right time. In our live deployment the room stayed inside the comfort band 99% of the time while the system ran itself.
Nobody, all day. It runs itself and tells the team only when something needs a look — in plain English, with the likely causes listed underneath.
The building keeps running. Data is kept on site and a local copy of the dashboard keeps working. The safety bounds are enforced locally, so the system stays inside them even offline.
Yes. The dashboard works on laptops, phones and wall screens — the status board is built to be left on a control-room TV.
A printable report per unit: running hours, how often the room stayed comfortable, average temperatures and fan speed. One click, ready for the monthly review.
Only the right people. Viewers can look; managers can operate; admins set schedules and limits. Every action is recorded — the dashboard keeps a log of who changed what, and when.
Yes. It is live at Candor TechSpace in Gurugram — 33 AHUs across two towers — and at Westin Hotels, with 12+ months of continuous autonomous operation and zero comfort or safety incidents.
Kaivalya Innovations is a physical-AI company from Gurugram, India, founded in 2024. We started with a simple observation: buildings are full of intelligence that only recommends — dashboards, alerts, reports — while the equipment still waits for a human to act. We build the layer that acts.
SKW EnergyOS is the result: an AI system that autonomously operates commercial HVAC on top of the BMS a building already owns — perceiving live state, deciding from physics, and writing setpoints in a closed loop, always inside hard safety bounds a human can override.
We earn autonomy in stages, never assume it. Our systems run offices and hotels for some of the largest real-estate and building-technology names in the country, around the clock — for operators, the proof is uptime, not slideware.
We also write — on control loops, energy, and why civilisation's story is an energy story:
Where this goes
Built as a fleet-learning platform, so every building it runs is intended to make the next one smarter. The highest-stakes version is the AI data centre, where freed cooling power becomes usable compute.
Vision and roadmap — not a claim about today.
A walkthrough of the live 12-month deployment, what autonomous control would look like on your BMS, and the safety model in full. No obligation, no sales deck.