Deploying artificial intelligence (AI) across built environments requires adapting operational logic to the specific priorities of individual asset classes rather than applying uniform automation rules, according to a technology founder. While commercial office spaces and specialised facilities rely on similar structural hardware and sensors, operational objectives vary significantly depending on the purpose of the property, said Arfaz Khan, founder and chief technology officer of ARVIS.The startup was established by Khan, alongside co-founders Afnaz Khan and Muhammed Naseef, drawing on a team of engineering and product design professionals from India. Supported by Qatar Science & Technology Park (QSTP), the venture created its Adaptive Building Intelligence system to serve as an operational advisor for large-scale facilities, connecting directly with existing Building Management Systems, heating, ventilation, and air conditioning setups, and environmental sensors to collect operational performance data.Khan said building management systems process identical environmental data differently based on operational context. “A hospital and a commercial tower may have very different priorities, but underneath, you’re still dealing with buildings generating information and people trying to understand what that information means,” Khan explained. He noted that an office space experiencing low occupancy can prioritise energy efficiency, whereas healthcare facilities must maintain uninterrupted operational continuity regardless of power demand. “An office with low occupancy might be able to prioritise efficiency. A hospital may have to prioritise continuity. A critical facility might care about reliability above everything else. The data point can look exactly the same, while the right action can be completely different,” Khan said. Khan pointed out that developing effective building intelligence involves teaching algorithms why specific events matter within a given setting rather than simply detecting basic anomalies. “The hard part isn’t teaching a system to recognise what is happening. It’s teaching it why that particular situation matters in that particular building,” he stated. He emphasised that local testing in active commercial properties provides critical feedback that isolated laboratory settings or synthetic datasets cannot replicate.“With ARVIS, I can’t just take a dataset, build a model and assume I understand how a building works. I need to understand the systems, the people operating them, and what actually happens when something goes wrong,” Khan said. He explained that early conversations with facility operators altered the development of their technology, shifting focus from data aggregation towards decision-making clarity. “We had assumed the hard part was bringing a building’s data together into one place. Then we spoke to people who have spent years actually operating buildings, and they pushed back on a more important question: once you have all that information, what does someone actually need to know to make a better decision?” Khan recalled.