Every platform is “AI-powered.” Every system claims optimisation. Every dashboard promises intelligence. But how do you which solution will work for you (if any)?
Here are five questions to ask before deploying AI into your building portfolio.
1. Will It Deliver Real Wins — Or Just More Data?
Real-time appliance data sounds powerful. And it is.
But only if it leads to action.
Across commercial buildings, around 40% of energy consumption comes from plug loads — and a significant portion of that happens when no one is there. Monitors in standby. AV equipment left running in empty rooms. Water heaters boiling overnight.
Most traditional BMS miss this energy waste because its invisible at meter level.
AI has value when it moves beyond reporting and starts controlling — when it can:
• Identify patterns of non-use
• Differentiate active vs standby energy
• Automatically eliminate out-of-hours waste
• Surface anomalies that would otherwise remain invisible
In our own office, we suspected we were over-provisioned on desk space but we lacked reliable data to confirm it. Badge data didn’t reflect actual desk usage. Surveys quickly became outdated. Installing additional occupancy sensors felt disproportionate.
Instead, we used appliance-level occupancy signals from smart sockets. By analysing real usage patterns, we discovered peak simultaneous desk use rarely exceeded eight desks despite having provisioned for far more.
The result was a 77% reduction in office costs and £47,000 in annual savings, achieved by right-sizing space based on actual behaviour rather than assumptions.
AI delivers value when it enables intervention — not just insight.
The key question isn’t “Does this have AI, will it give me more insight?” It is, “Can it deliver measurable improvements without relying on constant human oversight?”
2. Is the Automation Adaptive — Or Static in Disguise?
Many legacy building systems rely on fixed schedules and rule-based controls.
Compare one month to another. Adjust time schedules. Hope occupancy hasn’t changed.
Hybrid working, flexible hours and fluctuating occupancy have introduced variability that static controls struggle to accommodate.
AI should not simply replicate rule-based automation in a new interface. It should continuously learn from real-world behaviour using live energy signatures to determine when spaces are genuinely in use.
There is a meaningful difference between:
• A rule that says “turn off at 7pm”
• A system that knows nobody is there.
For example, the plug load management system from measurable.energy, applies this logic at appliance level. The AI-powered smart sockets use real-time energy signatures to distinguish between active use and standby, automatically switching off non-essential loads when spaces are unoccupied — without relying on fixed schedules or intrusive sensors.
This kind of adaptive control reduces manual intervention while responding dynamically to how buildings are actually used.
3. Are You Replacing Infrastructure — Or Adding Intelligent Integration?
One of the most common assumptions around AI in buildings is that it requires major system change.
Rip out the BMS.
Install new sensors everywhere.
Redesign infrastructure.
In reality, some of the fastest returns come from the simplest interventions.
Energy intelligence does not always need to sit at the centre of the building system. Sometimes it’s more powerful at the edge — at the appliance level — where energy waste is visible and introduced incrementally.
Before investing heavily in upgrades, consider:
• Can intelligence be layered onto what already exists?
• Does it offer appliance-level monitoring and control?
• Can it unlock rapid efficiencies without disrupting core building systems?
In many cases, the fastest ROI does not come from replacing systems, it comes from making them smarter.
4. Can You Prove ROI with Confidence?
Sustainability leads are under increasing pressure to prove ROI of their projects, competing for investment against other operational priorities.
Static baselining struggle to account for behavioural and operational variability, making it difficult to link reported savings directly to financial outcomes.
If savings cannot be correlated with utility data and operational change, claims of optimisation risk losing credibility, particularly as reporting requirements tighten under frameworks such as SECR and ESOS.
So, look for AI-powered solutions that can provide:
• Granular measurement at appliance level
• Clear attribution of avoided consumption
• Transparent reporting that correlates with real spend
• Carbon reductions that are directly traceable.
Confidence does not come from dashboards.
It comes from measurable outcomes.
And in a climate where sustainability teams must justify investment decisions, that confidence matters more than ever.
5. What About Security?
A growing concern — particularly in finance, healthcare, and manufacturing — is cyber risk.
If AI requires deep integration into core infrastructure, it raises legitimate questions about:
• Network security
• Data governance
• Operational disruption.
As building systems become increasingly connected, cyber security and data governance cannot be secondary considerations.
Look for AI-powered solutions that:
• Integrate into your current operational infrastructure
• Demonstrate secure architecture
• Offer controlled access and clear data handling practices.
If AI introduces operational risk, the ROI equation changes.
A Question of Control
Our recent webinar summarised the principle simply:
• Real-time data should lead to quick wins
• Automation should evolve beyond static rules
• Intelligence should enhance, not disrupt
• ROI should be provable, not assumed.
AI in buildings is not about sophistication for its own sake.
It’s about control.
Control of waste.
Control of cost.
Control of outcomes.
The organisations that benefit most from AI in building management are not those chasing innovation headlines.
They are the ones asking better questions before they start.