The Marketing vs. The Operator Experience
By mid-2025, every major PM platform offers some flavor of AI leasing assistant — AppFolio's Lisa, EliseAI, Funnel Leasing's chatbot, RealPage's IRIS, and several independent vendors. The marketing claims tend to land in the same range: 25-35% reduction in leasing-team workload, faster lead response, higher conversion. The operator reality is messier.
Across firms we have spoken with that have run AI leasing for 12-18 months, the consistent picture is: real wins on the boring stuff (after-hours inquiry handling, tour scheduling, FAQ responses) and disappointing performance on the things that actually drive conversions (handling objections, navigating co-signer questions, working with voucher applicants).
Where the ROI Is Real
- After-hours inquiry response: 35-45% of inbound leasing inquiries arrive between 6pm and 8am. An AI agent that responds within 60 seconds at 11pm closes 2-3x more of those leads than the same inquiry waiting until 9am for a human callback.
- Tour scheduling: Calendar back-and-forth is genuinely solved. AI agents can move a lead from "interested" to "tour booked" in 4-8 messages without a human touching the conversation.
- FAQ deflection: Pet policies, parking, application requirements, deposit amounts — these account for roughly 40% of leasing-team inbound minutes and are now largely automated.
Where AI Is Still Underperforming
- Objection handling: When a prospect says "the rent is too high," the AI's default response is some version of "the property offers great value." Human leasing agents pivot to concessions, term variations, or alternate units in the portfolio. AI agents rarely do this well in 2025.
- Voucher and SOI conversations: Few AI tools are trained well on Section 8 program mechanics. Operators in SOI-protected jurisdictions need to be especially careful — an AI that handles voucher inquiries poorly is a fair-housing risk.
- Edge-case applicants: Self-employed prospects, recent immigrants without US credit history, retirees with assets but low income — AI tools default to "thank you for your interest, here are our standard requirements," which is exactly the wrong response.
The Pricing Has Gotten Predatory
AI leasing pricing varies enormously. Native modules in your PMS run $0.50-$1.50 per unit per month. Standalone vendors charge $3-$8 per unit per month, sometimes $200-$400 per property as a flat monthly fee. For a 100-unit property, that is the equivalent cost of a part-time leasing assistant. Several vendors have moved to performance pricing — paying per qualified tour or per lease — which is the right structure if you can negotiate it.
The Implementation Mistakes
Three implementation mistakes show up repeatedly:
- Letting the AI handle the entire funnel. The handoff to a human at the right moment is what makes the system work. Most teams set the handoff too late.
- Not training the AI on your specific portfolio. Out-of-the-box AI agents do not know that the south-facing units in Building 3 have a view premium. Loading property-specific knowledge into the system is the difference between mediocre and good.
- Not auditing the conversations. A weekly review of 10-20 AI conversations is the only way to catch hallucinations, awkward phrasing, or compliance issues before they accumulate.
The 2026 Outlook
The capability gap on objection handling is closing fast — voice agents from several vendors materially improved between late 2024 and early 2026. Within 12-18 months, expect AI leasing agents to handle 60-70% of the funnel end to end on standard applicants, with human escalation reserved for genuinely complex cases. The firms making the investment now are building the prompt libraries and audit workflows that will be hard to replicate later.