The amount of growth experienced by the PropTech industry in the last decade has been stunning, to say the least. With more than 1,000 new companies vying for the attention of CRE industry giants it seems a new solution for an antiquated, manual process pops up every quarter. The latest being AI-powered tools with the ability to automate and anticipate tasks in historically time-consuming processes.
AI tools have come a long way and there’s still room to grow, but Daniel Gonzales recently wondered in a post on LexBlog whether automation of commercial lease management is a technological advancement of the future or the present.
Automation can be used to describe a very wide range of technologies. But generally speaking, automation occurs when a process or a task, previously performed by a human, is performed by a machine or device with the intention of minimizing human interaction. Automation technology is predictive in nature, namely predetermining decision making, relationships, and actions.
In CRE, innovations like automation stand to have tremendous impact in a process like lease abstraction or portfolio management which previously was done manually at great time-expense to an organization. Many commercial portfolio documents, like a lease, are filled with “hard terms” like rent amount, common area maintenance (CAM) charges, a commencement date, etc.
These terms, which are objective and quantifiable, are great building blocks for an artificial intelligence framework. In this sense, AI tools are very useful and very much here, today. Just look at some of the complex lease abstraction capabilities of a tool like Prophia. But commercial lease abstraction presents some challenges for which the technology just isn’t ready and there isn’t a one-size-fits all tech solution.
In addition to some predictable, quantifiable terms in every commercial lease, leases also contain what tech experts refer to as “dependencies”, terms subject to change based on the subjectivity of a deal and therefore difficult for automation to accurately account for. Daniel Gonzales gives a pertinent example:
“Consider the commencement date: if the commencement date changes, what other parts of the lease also need to change? In other words, how many other parts of a lease are dependent on the commencement date? Expiration, option notices, free rent schedules and letter of credit burn-downs to name a few.”
The negotiation period of any lease poses a difficult scenario for AI’s ability to predict or anticipate the nuances of a commercial real estate deal. If some of these “hard” terms rely on final negotiations, suddenly, they become “soft” terms, and many (if not all) terms in an initial draft will need to change based on logic “in the heads of the legal team”, as LeasePilot states.
This is where some of the current limitations lie. And if AI wants to grow with the increasingly complex needs of commercial landlords, management, and legal teams, it will need to learn to interpret context with the accuracy of a human.
So if CRE deals and documents are becoming more complex and the job of a property management team more tedious, how can AI grow into a technology that is sophisticated enough to accurately capture and predict data with nuance? Many experts believe the solution is context-aware AI.
Contextual AI is very complex and still in nascent stages for everyday use. That’s not to say, however, that there aren’t already some examples of contextual AI changing tasks we’ve been doing manually as a society for the past 100 years. The automobile is a great example. Self-driving cars are one of the first instances of “context-aware” AI, that is, the machine in the car is aware of all of the same elements as a human driver i.e. the road, any passengers, potential danger, the state of the car, etc. The awareness of these variables and the ability to predict outcomes and react, is considered a fully Contextual AI System.
So how does this relate to CRE? As you can imagine, if AI is sophisticated enough to make a decision like gently applying brakes in a way that maintains control of a vehicle and prevents a crash, then in a use case like lease extraction, contextual AI will relate terms like the commencement date, effective date, or termination date to predict an outcome: using one and two to calculate three.
While fully integrated Contextual AI Systems are considered a technology still in development stages, Prophia’s intelligent lease abstraction capabilities are always improving with the help of human expertise. Using a parallel tagger, Prophia is able to learn from the human tagging process, picking up on patterns in the data and tagging key terms in lease documents like a human.
Using, again, our example of the effective date, if a human tags one data point with the name “effective date” in one document, parallel tagger will take in the context of that tag, such as the name of that tag, what the tag looks like, and the tag’s context and it will tag all subsequent effective dates in other documents within the data room.
This awareness of the tag’s context allows Prophia’s AI to apply what it has learned to a new tag even if the human tagger tags a term the AI hasn’t seen before. This is just one example of Prophia’s ability to understand increasingly complex data so property management teams, lease administrators, lawyers, and asset managers can all work from the same accurate database without wasting time and effort.
Find out more about how Prophia works, today.