Results

The Building that Learned to Read the Room

At a glance

Why the smartest fix wasn't teaching people about the heating,

it was teaching the heating about people.

100% CAPEX recovered in year one

%

100% CAPEX recovered
in year one

25% Reduction in energy use across both gas and electricity

%

25% reduction in energy use
across both gas and electricity

74% Fewer manual radiator adjustments

%

74% fewer manual radiator adjustments

The Challenge

Quantum Spark was given a known problem in a 1960s commercial office building and a clear challenge: prove that intelligent control could solve it with a positive return on investment in under three years.

Frustration: 

Staff reported that rooms swung between feeling freezing cold or unbearably hot.

Waste:

Energy bills kept climbing despite new radiators with thermostatic valves.

The Test

Could a relatively modest intervention deliver enough measurable savings to pay for itself inside three years?

It paid back in year one.

"Automation follows a rulebook someone else wrote, whereas an AI agent is given a goal to achieve. It works out for itself, continuously, how best to reach the goal as the seasons and the people using it keep changing."

Rich Lansdowne
Director, Quantum Spark

A diagram contrasting the traditional automation method of devising schedules and automating them with the Agentic approach of setting a goal. Thus allowing the system to determine how to achieve the goal and how to measure its own performance.

Review and Strategy

A view of the outside of a 1960s office building

A winter site visit revealed an immediate contradiction: unoccupied rooms were being heated needlessly, while staff in some offices were shivering in the morning. Radiator valves featured traditional 1–5 dials—arbitrary numbers that led to chaotic settings. Colleagues at neighbouring desks often had set their valves to opposite extremes, turning the heating into a tug-of-war. The underlying problem was clear: the building was running to a fixed clock rather than human need, while each radiator was effectively controlled by the last person to touch it.

 

The strategy was simple: bring all the radiators under an intelligent, room-by-room control system that would learn to optimise for the needs of the people and their occupancy habits as well as facilities bookings, while compensating for the effects of the weather.

Execution

First...

Establish the Baseline

Before changing controls, we installed energy monitoring and started to capture weather data that would document our starting point while allowing us to understand how each room responded to the physical world.

Then...

Gather Human Data

We deployed off-the-shelf smart radiator valves and occupancy sensors, while logging every manual interaction. This allowed us to learn how each room responded to weather, how people actually used the space, and how their interventions changed conditions. Together, that created the knowledge base the AI Agent needed. With that data, we could link actions to reactions, learn habits, and detect and anticipate issues before they became problems.

Finally...

Deploy Agentic AI

We then built “The Agent,” an intelligent control system that tuned each room independently. Rather than heating the building to a fixed timetable, the system aligned each room with how it was actually used, the needs of the people inside it and the changing effects of the weather.

The Unexpected Bonus

A smart wireless radiator valve attached to a radiator

As would be expected, controlling the new radiator valves reduced the gas usage, but electricity consumption fell by almost the exact same proportion. Why? Staff had previously coped with cold mornings by running personal electric space heaters under their desks. Because The Agent ensured rooms were comfortable before people arrived, those space heaters were simply never turned on. A significant source of electrical waste disappeared without asking anyone to change their behaviour.

 

The system quietly earned people's trust without asking them to think about it.

The Takeaway

The best technology is invisible

Quantum Spark didn't overhaul the building with expensive infrastructure or mandate new rules for staff. We gave off-the-shelf hardware an intelligent brain. Employees don't care about algorithms or machine learning; they simply walk into a comfortable room, focus on their work, and leave the heating alone.

 

Want to explore what this approach could do for your estate?

 

Quantum Spark Logo

Delivering Your Vision