After investigating dozens of proptech vendors claiming to offer “agentic AI,” here’s what I found:
Most companies selling “autonomous AI agents” for property management are actually selling fancy chatbots with calendar access. Think of it like this: they’re selling a smart assistant that can schedule appointments, not a replacement for your property manager.
What AI Can Actually Do Today
- Handle 60-80% of basic maintenance requests (like "my AC isn't working")
- Match work orders to the right vendor automatically
- Read and sort documents with 95% accuracy
What AI Can't Do
(Despite the Sales Pitch)
- Diagnose why that AC isn't working without human help. See Note 1 bottom
- Make legal decisions about rent adjustments
- Think through complex, multi-step problems independently
The term “agentic AI” implies these systems can reason and plan like humans. In reality, they’re more like really smart templates that follow pre-programmed rules. It’s the difference between a GPS that reroutes around traffic (helpful!) and a self-driving car that doesn’t exist yet.
Yes, there are increasingly effective agentic cars, though we’re still in a transitional phase.
Bottom line for property managers
AI tools can absolutely save you time on repetitive tasks. But any vendor promising their AI will “think” for you or make complex decisions autonomously? They’re selling you tomorrow’s technology at today’s prices.
Ask vendors for specific examples. If they can’t explain it simply, they probably don’t understand it either.
Watch for common exaggerations that follow predictable patterns. The FTC’s “Operation AI Comply” in September 2024 targeted five companies for deceptive AI claims, while the SEC charged two investment advisors $400,000 for “AI washing” in March 2024. Failed initiatives, such as DoNotPay’s “robot lawyer” ($193,000 FTC settlement) and various business opportunity schemes that exploit AI hype, illustrate the prevalence of misleading claims. Red flags include promises of “revolutionary” transformation, guaranteed ROI without risk disclosure, and claims about eliminating human oversight entirely.
Key Red Flags to Watch For
When evaluating AI vendors at conferences, watch for these warning signs:
Claims like "10.46% profit boost" without third-party validation or detailed methodology
Companies claiming multifamily expertise when they actually serve different markets
Most current solutions are enhanced automation, not true autonomous agents
Realistic improvements are 5-15%, not complete transformations
Terms like "bridging data intelligence" often mask implementation complexity
Legitimate vendors offer trials and staged implementations
Credible companies have verifiable client testimonials
Questionsto Ask Vendors
Before committing to any AI solution, demand answers to:
- What specific operational metrics improved for your named clients?
- Can you provide third-party audits of your performance claims?
- How long did implementation take for similar-sized portfolios?
- What percentage of implementations failed or were abandoned?
- What ongoing costs exist beyond initial licensing?
- How do you handle integration with legacy property management systems?
- What happens to our data if we terminate the relationship?
- What level of technical expertise is required from our staff?
#PropTech #MultifamilyHousing #AIReality #PropertyManagement #TechTruth
NOTE 1:
Agentic AI could potentially diagnose many AC issues without human help, but with important limitations. Here’s what’s currently possible and what’s not:
What Agentic AI Could Do:
- Analyze sensor data: If connected to IoT sensors monitoring temperature, pressure, electrical current, vibration, and refrigerant levels, AI could identify patterns indicating specific failures
- Process error codes: Modern HVAC systems generate diagnostic codes that AI could interpret instantly
- Review historical patterns: AI could analyze maintenance logs, past failures, and performance trends to predict likely causes
- Cross-reference symptoms: Match multiple data points (age, model, symptoms, weather conditions) against vast databases of known issues
- Guide robotic inspection: Direct drones or robots with cameras to inspect units for visible issues like ice buildup, damaged fins, or leaks
Current Limitations:
- Physical inspection needs: Can’t check for refrigerant leaks without sensors, physically see ice formation, or smell burning components
- Manual testing required: Can’t physically test capacitors, check electrical connections, or clean components
- Sensor dependency: Only as good as the sensors installed – many apartment ACs lack comprehensive IoT monitoring
- Complex diagnostics: Some issues require hands-on testing with specialized tools (manifold gauges, multimeters)
In Multifamily Context: For apartment complexes, AI-powered building management systems are increasingly able to:
- Predict failures before they occur based on performance degradation
- Automatically dispatch work orders with probable diagnoses
- Optimize maintenance schedules across hundreds of units
- Identify building-wide patterns (like multiple apartments failing similarly)
The technology exists, but it requires a significant amount of sensor infrastructure that most current apartment buildings lack.
