Guest blog by, Rouben Alchoujian, Principal, Praedium Consulting

Commercial Real Estate’s slow tech adoption is a known fact, yet, things are changing, especially in corporate occupier portfolio management. CRE (Corporate Real Estate) is more about data and less about physical structures, so it is more likely to be open to innovation and new tech adoption than more traditional sectors of Commercial Real Estate. Artificial Intelligence, one of the most exciting and powerful tech disruptors, is already being used in legal documents review and, in particular, in abstracting Real Estate leases. For the past year this concept has been gaining popularity, and now appears to be moving from ‘early adopters’ to ‘early majority’ on the evolutionary curve. The benefits of its implementation are significant, and what’s equally important for deployment decisions, easily measurable.

 Background

Lease abstraction is a fundamental element of managing corporate occupier leased portfolio data – a CRE function known as Lease Administration. Understandably, completeness and accuracy when abstracting a lease are extremely important as errors can be very costly. Having a lease abstract contain incorrect information, or miss critical data, is similar to constructing a building on a faulty foundation. Therefore, lease abstraction is a very rigorous process, which typically includes up to 3 layers of review after the abstractor has done their work. And, sure enough, the team needs have adequate training, skill set and knowledge which typically comes with years of relevant experience.

With Artificial Intelligence, years of experience are being substituted by the machine learning (ML) process. Once the program is trained, it can generate a lease abstract in seconds – compare that to typical 3 – 5 hours required for a manual process.

 Product

Given the tremendous opportunity this generates, there are an increasing number of tools now available. Notable examples, include, eBrevia (part of Donnelley Financial Solutions), DealSumm, Diligen, Kira Systems and Leverton (recently acquired by MRI Software)

All these tools use machine learning as their core operational model and follow a similar pattern. At a glance, the legal (and Real Estate) team who identifies the most common lease provisions that would be required for lease abstraction. Then, the algorithms are designed, and put to work for each of these provisions. Finally, the actual machine learning process takes place, and includes human feedback and subsequent self-correction. After years of utilization and ironing out the glitches, lease abstraction tools are now deployed in growing numbers in North America and globally.

How It Works

The principles of AI (machine learning in particular) are applied to teach the machine to recognize the relevant lease provisions in the lease agreement, and then display them for the user. Even though most tools have been already trained on hundreds of most commonly used lease provisions, the user can still add their own if they need to.

With minimal user training it’s easy to move through the steps of the automated abstraction process. First, the leases get uploaded in the tool, and then OCR-ed (converted to machine readable format). Next, the appropriate model is chosen by the user, which basically means selecting from hundreds pre-trained lease provisions (‘models’). This step only needs to be done once (for a particular Real Estate portfolio), which allows any new lease in that portfolio to be abstracted based on the same template. Finally, the abstraction is launched, and the output generated in a convenient format.

All the systems have a fairly easy UI (user interfaces) allowing for easy upload of leases, navigation between the lease records and portfolios, running reports, as well as tracking the abstraction status and the progress.

Users

Multimarket corporate occupiers (companies with multiple locations) have different requirements for the lease abstraction volume. Those with a few dozen locations would only need to abstract one, or a few, leases a year which generally doesn’t justify the effort and cost of automation.  For larger occupiers with hundreds or thousands of leases, on the other hand, it makes a lot of sense.

In addition to the corporate occupier, the user groups that benefit most from lease abstraction automation are those who provide the services for them. These are CRE service providers, Lease Administration service providers, and, most certainly, specialized Lease Abstraction service providers. With the abstraction volumes of thousands of leases a year, there is significant potential for cost savings. It comes as no surprise that many of these companies have been looking into deployment of AI for lease abstraction.

 Benefits

Abstracting a lease takes an average of 4 hours and costs approximately $150 US, e.g. if outsourced to a lease abstraction vendor. Abstracting a lease using an AI-based tool takes an average of 5 minutes (out of which 4 minutes and 58 seconds are for upload and setup – still needs to be done by a human!) and can cost as low as $7 US. There are various pricing structures and models, some of which include quality control procedures involving human review.  However, even after factoring these in, the savings are so impressive that AI based lease abstraction can be widely used as a strong showcase of the benefit of AI deployment.

 Rouben Alchoujian, Principal, Praedium Consulting