Asset Management Bots Herald New Age of Predictive Analytics

Dottid’s new custom-built AI models for asset management, unveiled at CREtech this week, create a new form of over-the-horizon benchmarking that dramatically speeds the calculations now done by analysts and brings the emerging technology to the doorstep of bot-powered predictive analytics.

According to CEO Kyle Waldrep, the models are designed not only to speed real estate decisions, but, ultimately, to “get in the face” of decision-makers and push them in the right direction.

“The near-term use case is predictive analytics. We want to know what’s going to happen—pick your timeline, in six or 12 months—over the horizon and we want to benchmark against that,” Waldrep told Globe St.

“The AI should be a call to action. It should be a prompt to the person in charge to take a look at something, to make a better decision, to read the numbers—it should be in your face enough to where you have to go do something with it,” he said.

Dottid is custom-building models for individual clients around their data sets with the security to protect that proprietary data, while at the same time layering it with an “API infrastructure” that allows the model to be “cross-pollinated” with partnership data.

“AI should interact with each customer’s data set and only their data set. We’ve taken the time to build that internally and secure it,” he said. “We’ve also built a large API infrastructure that allows us to bring our partnership data into our modeling sets for our customers.”

“That’s where the fun begins, because you can start cross-pollinating the data from an API perspective inside of those integrations and you can really start creating forward-looking benchmarking, new metrics and new calculations,” he added.

The Dottid CEO gave us an example of how the bot-aided scenario planning and calculations will play out in client conversations with new AI helpmates.

“Let’s say I’ve got an office asset and I want to know, what’s my down time in my vacant suites? I can ask my model how long has my suite been vacant, what’s my down time? I also want to know lost revenue. So, based on the last tenant in the suite versus no tenant in the suite, how much lost revenue have I had in that down time?” he explained.

“You’ll also want to know how all of this is affecting the value of the asset. You’ll start to perform calculations out of the AI instead of having an analyst or a group of people [do it] and it will pull forward the basic metrics so that you can reach a decision in a much quicker fashion,” Waldrep said.

Waldrep described the process of training the bots. Dottid’s CTO and VP Engineering were the key trainers—and their primary focus was on coaching the bots to understand CRE lingo.

“They’re training it on spelling, understanding the use of words, understanding the definitions of words that are common in real estate,” he said. “On the front end, we’re training the model to work with a customer who is real estate-centric and to respond to prompts that [are framed] in the customer’s lingo.”

The biggest CRE speed bump that AI models will overcome in the near term is the amount of time spent creating and generating reports in Excel and Argus, among other programs—along with a lot of the analyst jobs that are needed for those tasks.

“What we’re creating is going to add a lot of efficiency for people. AI can give you specific answers instead of spending a lot of time creating and generating reports,” Waldrep said.

So, who gets replaced by the bots?

“Where AI is heading is the optimization of the work you can get from people. If you optimize what you can get, then you’ll [need] fewer of them on the job,” Waldrep told us.

“It’s hard to know how far up the ladder it’ll go,” he added. “I don’t think it can replace brokers. There’s a lot of innate knowledge that brokers have—I don’t want to call it irreplaceable—but it will be much harder to replace.”

Will the bots get smart enough to make the big CRE decisions?

“I don’t know if it will make decisions, but it’ll make suggestions, because that’s what it should be doing,” Waldrep said.

“AI should be taking these large data sets and providing the calculations, and [then] it should interact with humans—the person that knows the real estate best, that knows the model best. That person should evaluate this just like they would any scenario or planning modeling in any other analysis,” he said.

Until the person who knows the real estate best is replaced by a bot? Waldrep conceded that the road to AI eventually leads to automation.

“Yes, the next logical step is automation, but I still think we’ve got a way to go to get that,” he said.

In the meantime, these AI co-pilots will offer a path to faster and more informed decision-making by CRE professionals, Dottid’s CEO told us.

“Specific answers lead to speed, speed leads to quicker decision making. If you have the right data, you can make the right decisions and, hopefully, you’re off to the races and running your real estate better,” he said.