On April 23, 2025, Charles Fisher and Blake Shaffer, CFA both from JLL, visited the Vault to discuss AI in Real Estate. Fisher leads Risk Analytics for America’s Risk Advisory business. In this role, he engages with global lenders and institutional owners to assess risk in their global portfolios using artificial intelligence and proprietary JLL technology. Shaffer is a Director in the Corporate Capital Markets and Net Lease practice group advising corporate clients on capital strategies to optimize the capitalization and monetization of their real estate portfolios.
Over the course of an hour, Fisher and Shaffer described how Commercial Real Estate (CRE) companies can deploy AI in their operations. They illustrated these applications with examples from their own operations and engagements. The first application they discussed was office automation, which could apply to most any industry. Specifically, they explained how AI interacts with third party brokers who may contact them about a property. The AI chatbot does a decent job of responding to most basic incoming questions. This saves a material number of hours for humans to do more value-added work. AI shines at such simple, well-defined tasks, Fisher explained. When non-standard questions are posed the chatbot will re-route the conversation to a human.
Shaffer paused at this point to emphasize the importance of data capture. AI efficiency depends on the volume of, and quality of, data fed into the system. At JLL, AI reads all customer notes created on the Salesforce CRM system. And every bid that is received by any JLL broker is fed into the data set for AI to consume. “Feeding the beast” is so important that some JLL managers are evaluated based on the timeliness of data recording by their teams.
Shaffer and Fisher went on to explain how AI facilitates faster identification of opportunities and risks. One opportunity cited was more efficient lead generation and prospecting. The potential client in this example is a commercial real estate tenant. Schaffer described a hypothetical where property leases have been rising rapidly in the area where the tenant is located. A rich data set, which contains lease rates, vacancies, and perhaps specifics on the particular tenant, would allow a CRE broker to approach the tenant with curated locations, well before the end of the current lease. Otherwise, the tenant would potentially be at the mercy of the current landlord when the current lease expired.
The duo from JLL claimed that its AI tools allow for more accurate and more timely forecasting of CRE fundamentals. Traditional forecasting requires manual data input and extensive calibration, which is time intensive and relatively inefficient. Assuming the raw material is high quality data, AI tools can absorb more data quickly and process it immediately. The result is superior model explainability and more expansive variable stress testing, accomplished on a platform that is scalable.
In another example, a client wanted to review its thesis that university student housing was a decent area of investment. The client turned to JLL to first analyze which universities were expected to grow faster than other institutions. JLL unleashed its AI systems to process data such as: academic reputation, tuition costs, financial aid availability, acceptance rates, freshman retention rates, salary and job outcomes, among other factors. Then, the analysis pivoted to the financial strength (or stress) of an institution. Among scores of other data, input included: credit ratings, endowment size, mix of in-state, out-of-state and international students, revenue diversity, as well as vulnerabilities to the current US Executive Branch’s education policies. Combining the analysis of growth and financial risk, JLL was able to offer its client insight into the risk of its current lending book as well as highlight new opportunities.
Having robust information built on scrubbed data is also valuable in processing new data. A client with 50,000 loans on their books asked JLL to help them clean and enrich the data on these real estate loans. The first step was addressing validation and verification, a task that would have been terribly expensive without the existing pool of information in JLL’s systems. In some cases, properties were not classified as clearly or accurately as they could have been. Shaffer and Fisher gave the example of a loan being made to a “Retail” client instead of to a “Major Retail Shopping Center.” The difference is important and JLL could quickly collate and organize properties for its client.
Shaffer and Fisher proceeded to explain how AI resources can enhance and simplify property monitoring and due diligence for any client, because of the scalability of the systems. JLL’s systems will automatically flag to clients’ assets that should be reviewed due to declining comparable property values or indications that risk is increasing. At the portfolio level, JLL can help clients understand where they may be over-exposed in terms of sector or geography.
And these insights can be delivered much more cheaply than was possible when such analysis was done on a project-by-project basis and required manually-intensive processes.
In short, the JLL duo explained that their company is providing their clients with better insights, on a timelier basis, and more cheaply than previously. The hero in the story is Artificial Intelligence, and its inherent analytical power. The foundation of this beautiful outcome (more valuable analysis at a lower cost) is clean, reliable data. Fisher and Shaffer emphasized that Garbage-In, Garbage-Out is the same problem it has always been. So, spend the time to feed the system good data sets and enjoy the benefits of powerful insights delivered more cheaply than before.