AI May Be The Only Way To Secure Legacy Construction Knowledge

InnovationAI May Be The Only Way To Secure Legacy Construction KnowledgeByShiva Dhawan,Forbes Councils Member.for Forbes Technology CouncilCOUNCIL POSTExpertise from Forbes Councils members, operated under license. Opinions expressed are those of the author. | Membership (fee-based)Jun 03, 2026, 08:15am EDTShiva Dhawan is the CEO and cofounder of Attentive.ai, the company behind Beam AI. gettyIn this moment, AI is not displacing preconstruction expertise; it may be the most viable path to preserving it.​​​The Wall Street Journal recently covered a wave of workers choosing to retire rather than adapt to AI. Just this week, companies like Meta began reassigning workers to AI roles. The implication is straightforward: In the current state of the industry, as AI arrives, standard labor demand often contracts.Construction is heading in the opposite direction. As the CEO and cofounder of Beam AI, an AI-powered estimating software, I’ve seen the same pattern play out repeatedly: Most experienced professionals aren’t leaving because of AI; they’re leaving anyway. Retirement in this industry is driven far more by demographics than by technology. And in this case, AI may be one of the only mechanisms capable of preserving what’s about to be lost—because what’s walking out the door right now is also decades of judgment, pattern recognition and decision-making experience.According to Spark, 41% of the construction workforce will retire by 2031, and 1 in 5 workers is already over 55. The industry is already operating under a $124 billion annual productivity gap driven by unfilled roles. But those figures still understate the real risk. This is a total knowledge system failure, one that shows up most clearly in preconstruction.​Estimating has always functioned as a kind of institutional memory, even if it’s never been labeled that way. A senior estimator doesn’t just price a job faster; they evaluate it differently. They carry a mental archive of what tends to go wrong: how scope...