On October 15, 2026, Honeywell officially introduced Forge Cognition, an agentic artificial intelligence and simulation architecture designed for commercial building operations. The platform was created through a technological collaboration with Nvidia and Dell Technologies. It aims to modernize facilities by deploying intelligent software agents that enhance operational efficiency without requiring owners to overhaul their existing building management systems.
The architecture connects directly to established on-premise building management deployments and evaluates operational telemetry in real time. Autonomous and semi-autonomous AI agents continuously process data feeds from previously disconnected building domains. Forge Cognition identifies cross-system patterns across heating, ventilation, and air conditioning, power grids, and physical security installations. Using this cross-domain visibility, the platform delivers intelligent alarm management that isolates critical emergencies and filters out low-priority alerts.
Predictive maintenance forms another core capability of the new Honeywell software. By assessing component wear and operating parameters, the system forecasts equipment failures before they cause disruptive outages for commercial tenants. Building operators retain precise governance over how much authority the agents possess. The platform allows managers to specify whether agents function in a purely advisory capacity, act semi-autonomously upon human confirmation, or execute control routines with full operational autonomy.
The focus on agentic software rather than physical automation reflects broader economic dynamics across commercial real estate operations. A research report published by Anthropic on October 14, 2026, provides detailed economic context for this trend. The study analyzed the intersection of artificial intelligence, robotics, and manual labor across physical operational environments, including the construction sector and ongoing facility management.
Anthropic found a significant divergence between theoretical technical capability and actual commercial viability in physical robotics. Modern AI and robotic hardware are technically capable of performing roughly three-quarters of all manual tasks, which account for approximately 34 percent of total working hours across the examined sectors. However, despite these technical capabilities, the financial case for deploying physical machines remains sharply constrained in day-to-day operations.
According to the Anthropic data, deploying physical robots is currently economically profitable for only 0.3 percent of physical tasks, representing approximately 300,000 jobs in the United States. High procurement expenses, complex hardware integration requirements, and costly ongoing error corrections continue to outweigh human labor costs. These economic constraints demonstrate why property technology capital and enterprise partnerships, such as Honeywell's work with Nvidia and Dell, remain focused on software agents and analytical workflows.