Sensor Data: The Key to Unlocking AI Investment Returns
According to the latest JLL report, the AI pilot rate in the real estate industry has surged from 5% to 92% in three years, but only 5% of surveyed companies report achieving a return on investment. This article points out that current AI pilots commonly lack critical data—the human experience within buildings. Through physical AI technologies such as thermal sensors, companies can capture space usage data without infringing on privacy, thereby optimizing office design, improving space utilization, and ultimately realizing the true value of AI investments.
Editor's note: This article is written by Honghao Deng, co-founder and CEO of Butlr. The views expressed in the article are solely those of the author.
When JLL released its report titled "AI Reality Check in Real Estate: 90% Pilot, Only 5% Achieve Full AI Goals," building operators paid close attention to the gap between AI applications and their return on investment. The report link can be found on the JLL website.
The report shows that building operators apply AI in three main ways, focusing on data-related workflows, portfolio optimization, and energy management. Notably, the proportion of real estate companies conducting AI pilots jumped from 5% to 92% within three years, but only 5% of respondents said they have actually benefited from their investments.
Given that AI is still in the pilot stage, it is not surprising that building operators show strong interest in the technology yet report low returns. However, the most noteworthy aspect of the report is how AI pilots are conducted—a key data set has been overlooked: the human experience within buildings.
Human interaction is crucial
Understanding human experience is central to creating collaborative and efficient workplaces. AI workplace data shows that irreplaceable interpersonal connections in the office hold significant value: employees can access physical environments such as laboratories, handle high-security data, communicate through conversation, and solve problems cross-functionally. These activities are difficult to replicate when employees are not working together on-site.
As the workplace continues to evolve, its design must revolve around these key high-impact interactions. This requires evaluation based on data that reflects how people use and experience spaces. There is no shortage of data tools showcasing various aspects of building management; data such as foot traffic and occupancy provide operators with baseline insights. However, understanding how people interact within spaces is the key to driving the data foundation for AI. Such intelligence comes from data collected by sensors—i.e., "Physical AI."
In the commercial real estate sector, Physical AI data sets can be obtained without infringing on personal privacy. This can be achieved through thermal-based sensors that detect movement and infer behavior based on body temperature while ensuring anonymity. Ideally, Physical AI data should be integrated with other facility management and real estate platforms to form a comprehensive view of the property.
Application examples of Physical AI
Here are two typical cases of Physical AI deployment in the workplace:
A global medical technology manufacturer.The company was considering leasing additional space due to the need for more administrative workstations. Before making an investment decision, they installed thermal sensors combining AI and body temperature sensing technology to understand the actual usage of laboratories and workspaces.
Without capturing any personal data, the sensors revealed that 30% of laboratory space was underutilized. This prompted the company to undergo an office redesign without incurring unnecessary life sciences fit-out costs—which, according to Cushman & Wakefield, average $846 per square foot. Additionally, the sensors helped the medical technology company identify the most frequently used instruments, enabling more efficient capital planning.
A global software provider.The company sought to improve its 15% workstation utilization rate. A mandatory return-to-office policy only increased office attendance to 45%. When the company reviewed AI sensor data, they found that employees would book large meeting rooms for focused work or phone calls and use the cafeteria for meetings in the late afternoon. They also observed a significant "swipe-and-leave" phenomenon—employees swiped in showing presence in the office but left quickly, completing most of their work from home.
The software company then reconfigured the office to suit employee needs. The new layout provided more small meeting spaces, flexible portable pods for private calls and focused work, and optimized cafeteria staffing schedules after the traditional lunch period. Within two weeks, space utilization increased to 72%. Beyond productivity gains, the facility management team improved operational efficiency and energy consumption based on aggregated data about human behavior collected by the sensors.
Implications for AI strategy
The rise of Physical AI is redefining how enterprises manage commercial real estate, transforming buildings from passive environments into active contributors to productivity, compliance, sustainability, and energy efficiency, while supporting smarter building operations decisions.
The key to realizing AI investment returns lies in the quality of the underlying data. This principle is not unfamiliar to many facility teams: they are integrating complex data sets covering nearly every aspect of building management, laying the right foundation for AI to function. However, it must be recognized that without the insights provided by Physical AI, these data sets will be incomplete.