2024 seems likely to be a year of rapid acceleration in the adoption of AI in multifamily. The question is how rapid and what kinds of tasks will it be doing for us? During this year's 20for20 interviews, the 20 leaders were asked to speculate a little on the direction AI is likely to take. What the conversations revealed is that most leaders view AI as something you buy from third-party vendors specializing in this area.
Multifamily currently sees AI much the same way it sees any other type of technology. That makes sense since most multifamily companies do not have the resources to explore and engineer their own AI solutions. However, some do, and the direction they're taking might indicate where this incredibly powerful technology is headed.
What happened last year
2023 was the first year the multifamily industry had access to generative AI, with the release of OpenAI's ChatGPT at the end of 2022 creating unprecedented opportunities. ChatGPT became the fastest adopted technology in history. The hype cycle was in full swing at industry conferences in the first half of 2023, with newly minted experts and increasingly exotic use cases taking center stage. The hype seemed to die down in the second half of the year, but that doesn't mean the ultimate progress of AI will be any less monumental.
AI will change most of what we do; we just don't yet know how. One clue to the adoption of generative AI came from discussions with these 20 executives. A small subset of interviewees was exploring, and in some cases testing the use of external large language models (LLMs) like OpenAI to address back-office processes. That includes, for example, enabling AI-enabled chat to handle basic helpdesk inquiries.
The more established AI use cases in multifamily are customer-facing, like digital leasing assistants, lead-nurturing and collections. All 20 respondents agreed that customer-facing capabilities should be purchased from specialist vendors, rather than created internally using LLMs. The only LLM-based use cases under consideration involved back-office functions. And that mirrors a trend observed with centralization over the last few years.
What this has in common with centralization
For the past three years, 20for20 has been monitoring the move towards a more centralized operating model. A surprising discovery last year was that, despite industry buzz about leasing centralization, it was the admin functions traditionally associated with assistant property managers that were being relocated offsite more rapidly.
This year's findings confirm this trend has not only continued but accelerated, suggesting it will soon be strange for companies to perform property admin tasks like accounting, collections, and deposit handling onsite. The evidence suggests that when companies start to make concrete plans to change operating models, they gravitate towards centralizing the activities that add the least value to the customer experience and generate the most friction.
We may be starting to see a similar thing with the adoption of generative AI. As organizations think about how it can have a more profound impact, the natural focus seems to be the back office. To date, the use of generative AI has been circumscribed, with individuals using apps like ChatGPT to write emails or draft website copy. Those use cases are helpful, but limited. The concurrent trends of admin centralization and automation may be starting to look like a path to real progress.
The path could continue the systematic relocation of admin tasks from properties to shared services, where they will increasingly be automated, with an expanding array of customized applications leveraging API access to systems like OpenAI.
The AI will eliminate the most repetitive and low-value tasks those central teams do. That will improve consistency and boost the productivity of the (fewer) individuals needed to manage processes for a far greater number of properties. That approach could radically redefine operating models, offering operators far greater leverage than they can currently achieve.
Brave new world
Customer-facing applications of AI are risky. They need a high degree of specialized training and the kinds of guard rails that only come through experience, which is why it makes sense for operators to buy such apps from point solution vendors.
The back office is a safer test bed for companies wanting to figure out how to leverage the incredible power of LLMs like OpenAI. It seems likely, then, that we will start to see companies finding administrative use cases for generative AI. But AI is not a technology that wants to stay in its lane: progress from initial, successful use cases may come quickly.
It's intriguing to think about what this might mean for multifamily operations. We tend to frame operations in terms of the tasks and roles that have always existed. This applies to “property admin,” which is most easily understood as the work normally done by assistant property managers.
But as the industry progresses toward centralization and automation, our definition of "admin tasks" may expand to encompass broader swathes of work. The path from centralization to automation may be the one that unleashes this most transformative of technologies.