Agentic artificial intelligence could become a powerful tool for rental housing, but only if owners and operators rethink how work is organized rather than simply layering on another technology platform. That was the message from Aditya Sanghvi, senior partner at McKinsey & Co. and a leader of the firm’s global real estate practice, during a session at Walker & Dunlop’s recent convention in Sun Valley, Idaho, which was shared as an August 12 Walker Webcast.
Sanghvi argued that rental housing is particularly well suited to agentic AI because property management consists of many simple but interconnected tasks involving residents, on-site teams and vendors. Those touchpoints can create gaps in execution, such as unresolved maintenance tickets, unanswered calls and missed follow-ups, which in turn weaken the resident experience.
He distinguished agentic AI from more familiar generative AI tools. While generative AI responds to human prompts, agentic AI is assigned a goal and then plans and executes the necessary steps, with the option to deploy multiple specialized agents working in coordination. In a maintenance workflow, for example, one agent might triage incoming requests, another could coordinate with vendors and a third could manage scheduling, while an additional agent analyzes recurring issues to inform better decisions over time.
Similar approaches could be applied to leasing conversions, renewals, rent collections and other repetitive processes. As agents encounter more scenarios, they can learn from past decisions and refine their performance. However, Sanghvi emphasized that AI will not replace human involvement. Humans remain critical to closing the loop on each agentic process, outperform AI on many tasks and provide accountability that automated systems cannot.
Despite the promise, Sanghvi noted that only a small share of companies report positive financial outcomes from AI today. He cited deployment complexity, poor data quality and organizational hurdles as key barriers. Real estate data is often fragmented across property management systems, spreadsheets and CRM platforms, and a Gartner study he referenced found that many companies have paused AI efforts due to data quality concerns. He added that leadership frequently treats AI as an IT initiative instead of taking direct ownership, and that change management is essential if employees are to adopt new tools.
To address these issues, Sanghvi advocated for a “rewired” approach that ties AI to measurable value, reimagines underlying workflows and embeds change management from the outset. He recommended focusing on discrete domains, such as lease renewals, with clear metrics and a defined data set. That focus can help teams clean and organize tenant information, test agentic AI in a contained environment and build feedback loops that steadily improve results. Strong data governance, guardrails and human oversight for significant decisions are also necessary.
Sanghvi concluded that rental housing’s repetitive, interconnected processes make it a natural fit for agentic AI, and that the companies that benefit most will be those that redesign work, improve data and create continuous learning loops around their operations.
