A thread in IFMA's All Member Forum this week pushed back on the idea that AI adoption can substitute for solid documentation practices. Donald Moore, FMP, opened the discussion arguing that predictive maintenance promises from software vendors ring hollow when the underlying asset records are incomplete, work-order notes are generic, and maintenance histories were never properly captured.
Adrian Merkel, CEO of Speedikon FM, added a useful counterpoint: don't let "build the foundation first" become an excuse to delay indefinitely. He noted that most organizations will never have a single clean, complete database — operational knowledge is spread across CMMS and ERP systems, BIM models, point clouds, building automation, and individual staff experience. The practical path, he argued, is to connect that distributed information rather than wait for one perfect system. AI can actually help with some of the groundwork — identifying duplicate records, classifying documents, extracting asset data, and flagging gaps in existing work orders. Start with critical assets, connect what already exists, and let every maintenance interaction improve the knowledge base over time.
For HTM and clinical engineering departments, this maps directly to a familiar tension: equipment data in the CMMS is only as reliable as what techs enter at the point of service. Departments leaning toward AI-assisted PM scheduling or predictive tools should audit their work-order completion rates and asset record quality before committing budget — partial data will produce partial, potentially misleading recommendations.
Source: https://engage.ifma.org/