𝗧𝗵𝗲 𝗖𝗘𝗢 𝗼𝗳 𝗠𝗮𝗿𝗰𝘂𝘀 & 𝗠𝗶𝗹𝗹𝗶𝗰𝗵𝗮𝗽 𝗷𝘂𝘀𝘁 𝘁𝗼𝗹𝗱 𝗪𝗮𝗹𝗹 𝗦𝘁𝗿𝗲𝗲𝘁 𝘁𝗵𝗮𝘁 𝗔𝗜 𝗶𝘀 “𝗳𝘂𝗹𝗹 𝗼𝗳 𝗲𝗿𝗿𝗼𝗿𝘀” 𝗼𝗻 𝗳𝗶𝗻𝗮𝗻𝗰𝗶𝗮𝗹 𝗺𝗼𝗱𝗲𝗹𝘀. He’s not wrong. He’s diagnosing the wrong problem. Last week, investors wiped nearly $12B from CBRE’s market cap in 48 hours. Not because earnings were bad. Revenue up 13% to $41B. Leasing up 14%. Property sales up 19%. The sell-off hit because Wall Street is pricing in a simple question: how much knowledge work inside these firms can AI replace? Every major brokerage CEO responded the same way. Our people are irreplaceable. Relationships matter. AI is a tool, not a threat. Then Hessam Nadji said something that stopped me: “When you try to apply AI to a financial model or a property overview or a marketing package, it’s full of errors.” I hear this constantly. Someone feeds AI a vague prompt, gets garbage back, and concludes the technology doesn’t work. That’s like blaming Excel for a bad underwriting model. Prompting is a skill. Structuring an AI workflow is a skill. Breaking complex tasks into steps the model can handle is a skill. Few in CRE have developed any of them. I build AI workflows for real estate analysis every week. T12 extraction, deal underwriting, market research. When the output is bad, the fix is almost never “AI can’t do this.” It’s “I structured the request wrong.” The skill gap is the real story. Not the technology gap. Now the bigger problem. Every CEO in that article promised to retrain their people. No layoffs. Just “evolving the talent pool.” What do the incentives say? These are public companies. They answer to shareholders. CBRE already told investors it will cut research costs 25% this year using AI. CoStar has eliminated roughly 500 roles through AI efficiencies. The moment a competitor automates 60% of the back office at lower cost, “retraining” becomes “restructuring.” Because the math will force their hand. Senior brokers with 30 years of relationships and pattern recognition? Safe. The three layers of analysts, marketers, and coordinators supporting them? Not Safe. So stop blaming AI for bad output. If you’re prompting it wrong, if you haven’t built the workflows, you don’t have an AI problem. You have a skill deficit. The CEOs promising to “retrain” their people? The retraining they’re describing is exactly this. Learning to structure AI workflows. Learning to prompt with precision. Learning to build systems that produce reliable output. The ones who figure it out will move faster with fewer people. The ones who don’t will keep calling AI “full of errors” while their competitors quietly prove otherwise. This technology changes by the month. The gap between those who build the skill and those who dismiss the tool is widening fast. Which side are you on? https://lnkd.in/e2jT_2sP