In almost every property office, there is now someone who has quietly become one of its most valuable people.
Not the best dealmaker, not the most senior head. The early adopter, working at a level of efficiency the rest of the team cannot match. For the moment, they are an edge.
That moment is shorter than most would think. In June, Anthropic released its most capable model yet to the public, a tier above anything it had sold before, one of several frontier tools any firm can now reach for the price of a subscription.
Within days, that same model was pulled, a reminder that these tools are rented, not owned, and can be switched off by someone other than you.
GETTING AHEAD OF AI
The intelligence and efficiency they are wielding is not scarce. It is the same intelligence every rival can buy by Friday, getting cheaper and better every month on a timetable none of us sets.
So the advantage is not the AI. It is being ahead on the AI, and being ahead is temporary by definition. The moment the whole industry adopts these tools, everyone is equally efficient, and an edge everyone shares is no edge at all. It is the new baseline.
Most leadership teams will respond to this in exactly the wrong way. Within the next six to 12 months, they will look at that one valuable person and most likely decide the answer is to make everyone like them.
The certifications will arrive; the AI-savvy badges; the mandatory training and the line in the budget. The trouble is not that this is hard. It is that it is easy and everyone is doing it at once.
Train the whole team and you match that one person. So does every competitor who trained theirs the same quarter. You have not built an advantage. You have funded the removal of the one you had and called it progress.
FINDING THE EDGE
So if proficiency levels out, where is the edge that lasts?
Not in the prompts and the skills, which every rival can buy and learn. It is in what is specific to your firm: the deals you have done, the sites you walked away from and why, the way you read a market nobody else covers as closely.
The work is to build that into your own workflows and integrations carefully, with high regard for security and for keeping your IP your own.
And the real prize is further out still. It is when those systems stop waiting to be asked and start working on their own, mining and generating across the business itself, every deal and document and the context around it.
Feed an AI that, and you have something no subscription can replicate.
RIGHT DIRECTION
The biggest names are already moving this way. JLL built its own platform to sit above a rotating cast of models rather than depend on any one of them. Cushman & Wakefield is understood to have rolled out Claude across the firm, built around its own data. Neither simply bought a tool off the shelf.
The limitation is not intelligence, it is context. Any real judgement on a deal is the product of dozens of factors at once, and an AI can only weigh them if it can see all of them.
Most of the time it cannot, because the information sits scattered: in spreadsheets; in old appraisals; in email chains; in the reasoning behind deals done and declined. Without that full picture, the tool is confined to the shallow end. Basic reporting, drafting, a first-pass sniff test on a site.
Useful, but not something you would trust to make a decision. It earns that trust only when it can draw on everything a person would, wherever that information happens to live. Short of that, you are asking it to make judgements with half the story.
That is the real work, and it is harder than buying a subscription or running a training day.
The race to make everyone AI-savvy ends with everyone arriving at the same place together.
The firms that pull ahead will be the ones that used the time to capture what is theirs alone and feed the machine the one thing no rival can rent.
