During Quest Forum Digital Event: JD Edwards Week, Ilan Zachar, CTO and Senior Vice President of Carr Properties, shared how Carr embraces innovation and made a move to integrate an AI Lease Abstraction system that had no integration with JD Edwards, but through understanding automation and JDE’s Orchestrator, was able to package a perfect solution. 

About Carr Properties

Privately held real estate company Carr Properties owns and manages high-quality office properties in greater Washington and Boston. Its 14 commercial properties total more than 4.4 million square feet, as well as three active development properties of 2.4 million square feet. Carr’s largest investors include a fund of institutional investors advised by J.P. Morgan Asset Management and Alony Hetz Properties and Investments.

In its presentation on AI Lease Abstraction and JDE Orchestrator, Carr Properties demonstrates how IT, which traditionally served as a support function, now generates value for its business.

With this in mind, Carr analyzed how it could function better with building leases, how its teams reviewed lease details, and how it could leverage third-party platforms to improve services overall. More specifically, Carr examined the benefits of implementing an AI solution to index a lease manually; looked at current reporting options lease details; as well as maintaining data integrity with JDE within our business and third-party systems.

AI Lease Abstraction and Orchestrator: Perfect Harmony

Carr partnered with DealSumm to leverage AI to make lease document abstracting, searching, and referencing efficient and accessible for all aspects of its business. DealSumm offered flexibility to build the right custom abstracts for Carr’s JDE approach. DealSumm configured it to where it would line up with the exact data in JDE Carr needed.

During pilot implementation, Carr wanted to find areas within its business where AI could learn and grow more accurately. It uploaded lease documentation into DealSumm’s interface. Then, through a feature called clustering, DealSumm’s AI engine used its algorithm to identify data connections within those documents.

Those data connections allowed Carr to provide search capability and deep-linking into the documents, with a folder structure organized by building, then tenant/customer, then lease documents.

In a real estate company that acquires buildings from different areas with leases written by different people, it can be really challenging to search across building, or search across portfolio if you’re just managing a bunch of PDFs.

-Aaron Altschuler, Head of IT for Carr Properties

DealSumm’s AI engine allows Carr to execute an accurate and deep search based on keywords, and filter through hundreds of thousands of pages of lease documents. For example, should Carr’s Asset Management Team need to turn off all water fountains in its properties, it can quickly and efficiently search existing lease agreements to ensure no clauses in those leases mandate water fountains be operational for a certain period of time, Altschuler said. “The team can use this system to search portfolio-wide for any property that has a water fountain or even provide greater search keywords to get what they need,” thus saving asset managers time and offering greater transparency.

Zachar said that one additional benefit is that where leases used to sit in a file cabinet or electronically on a server, now the abstract can directly link to the document. “There is a direct, one-to-one correlation from the item that you pull out of the abstract, compared to where it is noted in the lease itself,” he said.

Carr can now provide a reporting tool that allows business users to build custom reports. Each user can go in and add the columns they want, filters based on filtering requirements, then export that data in any common format.

Bridging the Gap

To bridge the gap between data in DealSumm and the lifeblood of Carr’s business, Carr uses JDE Orchestrator. Carr treats JDE as “the single source of truth,” meaning data enters one place manually and moves around between Deal Summ and JDE dynamically, Zachar said. To maintain data integrity, Carr built middleware, OrchAI and OrSync, and leverages Orchestrator to provide connections between the two products.

OrchAI and OrSync sit within Carr’s domain, interacting with the cloud-based DealSumm in two ways:

1. Carr takes its lease abstract data from Dealsumm and ingests it into our middleware through a JSON dump.
2. Data coming from JDE through our middleware back to DealSumm uses DealSumm’s newly defined post API, where Carr sends back posts to DealSumm on demand.

Carr interacts with Orchestrator by using middleware to create new items and validate whether items have changed. OrSync compares data in JDE, to what Carr receives from its third-party provider.

Business users interact with OrchAI as a nerve center that them to determine whether to bring data into JDE, or whether they’ve made updates that need to go back into DealSumm. OrchAI tracks all change history, version control, and storage of raw documentation.

Orchestrator adds a new lease by breaking the process up into steps called form requests. Each request bundles five – or in some cases, 20 – commands. Each request has a different function, which facilitates good error handling and troubleshooting. Examples of form requests include adding a new tenant number, creating a lease shell, logging detail notes, recurring billing, and sales overages.

DealSumm created value for Carr Properties in many ways. Using AI lease abstraction early on allowed Carr the maximum amount of time to train the tool to deliver precise, defined results, as it continues to gather millions of data points in its journey for ever greater accuracy.

Carr’s teams have traded in their time spent doing low-value work, like entering data, for high-value pursuits like analyzing data for potential weaknesses and identifying risks for customers. Carr improved critical date notification too, by structuring necessary data so it can run notifications, reports, and proactively communicate with customers.

In the future, Carr expects more improvement in the accuracy of AI abstraction. “We’re optimistic that we’re catching AI at the beginning of its maturity and growth, and the AI tools are only going to become faster, only going to become more accurate,” Altschuler said. “We’re seeing the only way to do that is to provide these tools the time and the data in order to grow and mature on their own.”

Carr plans to explore other document types it needs to summarize and index using DealSumm. “We’re seeing the relationship among JDE, Orchestrator, and DealSumm as a platform for any and all documents that we’re having trouble organizing, searching, filtering, and running reports against,” Altschuler said. “We can now create an efficient, end-to-end system for document management that becomes really smart and allows access to all users across the business.”