The Demo That Looked Ridiculous a Decade Ago Is Just Called AI Now

Every year, the company then known as DTZ gathered its top twelve hundred agents — essentially everyone who mattered in the business — for its annual symposium. And every year, as Chief Digital Officer and Chief Information Officer, I was asked to get up in front of that room and paint a picture of what was coming.

I took that assignment seriously. Maybe too seriously. Part of me was always trying to channel a junior Steve Jobs — less about the roadmap and the budget, more about the possibility. I wanted the room to feel the future before they could rationalize it away. I did skip the black turtleneck, though. For me it was always well-fitting jeans, boots, and a trendy sportcoat.

A Trash Bin Full of Stress Balls

One year, I walked out with a large bin and a simple rule. One stress ball, I told the room, represents ten gigabytes of data — about what you'd find sitting on the PC of a typical tenant rep broker. Every proposal, every pitch deck, every market survey, every spreadsheet and headshot and floor plan you've saved over the years. All of it, one ball. Then I started adding balls.

Balls for every broker in the room. Balls for every office. Balls for the CRM, the accounting systems, the property databases. The count climbed — a dozen, fifty, a hundred piled up while I kept talking. And here's the number that made it land: by 2016, the world was generating something on the order of 16 zettabytes of data a year. A zettabyte is a trillion gigabytes. At ten gigabytes a ball, you couldn't fit that many stress balls inside the planet.

And then I picked up the whole bin and dumped it out across the front of the stage. Thousands of stress balls, bouncing and rolling toward the front row. That was the point: the hundred I'd counted out by hand weren't the story. The flood was the story. Data wasn't just growing — it was multiplying, faster than any person could ever track by hand.

It got a laugh and a few wide eyes. But it set up the real question I wanted them sitting with: if this much information is pouring in every single day, what good is any of it if you can't actually use it in the moment you need it?

Which brings me to a later year — and an Amazon Alexa device.

The Conversation I Wanted to Have

My team and I spent real time building a demo around a single, simple scene. We put a real Amazon Alexa on the stage next to me, and up on the big screen we ran a mocked-up version of the device's display — her responses rendering on screen while her voice came back over the sound system. To the room, it looked and felt like I was simply having a conversation with my virtual assistant in front of a thousand people.

I imagined walking into my office before a meeting with one of the largest property owners in New York, and talking to that virtual assistant the way you'd talk to a sharp chief of staff:

And it went on from there. To a lot of people in that room, it sounded a little ridiculous. Some of them believed it was genuinely possible — but "possible" in the way flying cars are possible. Ten years out. A nice story to close a keynote, not something anyone would be using on a Tuesday afternoon.

I understood the skepticism. But I built the demo the way I did because I actually believed it — and I believed it for an unglamorous reason: the data already existed. Every single piece of it.

All the Data Was Already There

The dream wasn't science fiction. It was plumbing. Every source I'd need to answer that question was already sitting in plain sight:

• LinkedIn's Sales Navigator and Social Selling Index tools showed me the leaders in any peer group and the top voices on any given topic — and who was connected to me.
• The primary business press — the Wall Street Journal, Fortune, Forbes — was fully searchable for news, and Yahoo Finance had an excellent feed for company-level updates.
• EDGAR held every press release and filing you'd want for a public company.
• Our own CRM — the one we'd spent millions developing — knew who at our firm had touched the client, why, and which buildings and transactions we'd worked on together.

The data wasn't the problem. The problem was that answering one human question meant a person manually pulling threads from five or six systems that had no idea the others existed.

Start With the Question, Not the Warehouse

Here's the part that I think earned me some goodwill with my boss at the time. Early on, I'd made our CEO a promise: I would never be the kind of technology executive who shows up at his office every year pitching an eight-figure project that would take two years to deliver. I meant it. So I refused to solve this the way the industry usually did.

The default reflex was to spend multiple millions of dollars building a massive data warehouse — pour everything into one giant tank and hope value came out the other end. I'd seen that movie before. It almost never generated the value everyone promised at the outset, and it was exactly the kind of two-year, eight-figure adventure I'd sworn off.

Anyone who worked with me back then will remember the second one as something close to a religion. Instead of spending years and fortunes constructing warehouses, the exercise was simple: What are the ten questions that, if we could answer them instantly, would generate more value, grow revenue, and delight our clients?

Get the questions right, and the data models follow. Get seduced by the infrastructure first, and you tend to end up with an expensive answer to a question nobody asked.

Fast Forward Nearly a Decade

That demo was roughly a decade ago. The dream we mapped out on that stage is simply true today.

From what I can see in their public releases, Cushman & Wakefield is still building toward exactly this kind of capability — and so is every other serious real estate business. And it's bigger than our industry. All over the world, more and more people are asking the same question: how do we pull value out of all this available data faster and more dynamically than ever before?

I'll be honest about where I land on the hype. I think AI is overhyped in places. I think there's a genuinely difficult road ahead before anyone generates the kind of value implied by the valuations of today's major AI companies. That skepticism is real, and I hold it.

But I'll also admit I chuckle a little. What sounded ridiculous to a ballroom full of the industry's best agents a decade ago isn't a keynote fantasy anymore. It's a Tuesday afternoon.

The lesson holds up better than the technology, though. The reason that demo worked wasn't the Alexa device or the clever wiring behind it. It was the discipline of starting with a real human question and refusing to get distracted by the machinery. That's still the whole game.