Property Management: Cleaning & Maintenance in the Age of AI (Part 2: How AI Changes Things)

In Part 1, we explored the four main complexities of cleaning and maintenance management in property management. Now, let's dive into how AI is transforming the landscape by addressing these challenges.

Tackling the Complexity of Real-World Systems

The Challenge of Digital Twins

Ideally, we'd have a complete digital twin of each property, but creating this manually is impractical. This is where AI steps in.

AI-Powered Solutions

1. Video and Photo Analysis: AI's object recognition and classification capabilities have reached near-human levels, making data ingestion from visual sources highly effective. For example, an AI system can analyze a video walkthrough of a property and identify different types of flooring, appliance brands and models, and even potential maintenance issues like water damage or worn carpets.
2. Code and Regulation Integration: Large Language Models (LLMs) can provide superhuman insights into complex codes and regulations. For instance, if a property manager asks, "What are the fire safety requirements for a short-term rental in downtown San Francisco?", an AI system can quickly compile and summarize the relevant local, state, and federal regulations, even pointing out recent changes or conflicts between different codes.
3. Intelligent Decision Support: By combining property data with regulatory information, AI can offer nuanced advice. For example, when planning a renovation, the system might suggest, "Based on the property's age and local historic preservation codes, you'll need to use specific materials for the exterior. Here's a list of approved suppliers and estimated costs."

Enabling Extensive Customization

To be effective, a property management system must offer a vast range of customization options to capture the incredible complexity of user preferences. AI makes it easier to update this system and ensure it reflects users' wishes.

AI-Enabled Customization

Imagine a property manager says, "I want to use Cleaning Company A for all properties except the lakeside cabins, where I prefer Company B. But if either cancels, try to get an independent cleaner with at least a 4.5-star rating. Oh, and for the penthouse apartments, always send a team of two."

An AI system can:

1. Interpret this complex set of preferences from natural language.
2. Set up the appropriate rules in the backend system.
3. Create workflows to handle exceptions (like cancellations).
4. Continuously refine these rules based on feedback and outcomes.

This level of customization would be overwhelming with traditional interfaces, but AI makes it manageable and user-friendly.

Managing Probabilistic Events

AI Solutions for Probabilistic Management

1. Predictive Analytics: AI can forecast the likelihood of events. For example, it might determine that Cleaner X has a 85% chance of completing the job on time based on their history, current weather conditions, and traffic patterns.
2. Intelligent Scheduling: Using these predictions, the system might automatically schedule a backup cleaner for high-priority properties when the primary cleaner's reliability score drops below a certain threshold.
3. Dynamic Resource Allocation: If a maintenance task usually takes 2 hours, but the AI predicts a 30% chance it will take 4 hours (based on the property's history and the specific issue), it can adjust the schedule accordingly, potentially booking the next appointment with a larger time buffer.

Handling Fuzzy Inputs

AI Approaches to Fuzzy Inputs

1. Natural Language Processing: If a cleaner texts, "Can't make it tmrw, car trouble," the AI can interpret this, mark the cleaner as unavailable, and trigger the workflow for finding a replacement.
2. Image and Video Analysis: The AI can compare before-and-after photos of a cleaned room, identifying missed areas or quality issues that need addressing.
3. Sentiment Analysis: By analyzing guest reviews, the AI might notice a trend of comments about "musty smells" in certain units, prompting an investigation into potential moisture issues.
4. Contextual Understanding: If multiple guests complain about noise during their stay, the AI can cross-reference this with local event calendars or construction permits to identify potential causes and suggest solutions.

The Ideal AI-Powered Property Management System

An ideal system for cleaning and maintenance management would combine robust software with advanced AI capabilities:

1. Comprehensive Data Management: A strong foundation of customization options and workflows.
2. Smart, Conversational Interface: Allowing users to interact naturally with the complex system.
3. Intelligent Data Processing: Ability to understand and act on various inputs.
4. Learning and Improvement: Continuous refinement based on outcomes and feedback.
5. Human Oversight: Allowing for human intervention when needed.

AI Based Property Operations are Ready Today

With current AI technologies, automation of up to 99% of cleaning and maintenance management tasks is theoretically possible. The challenge now lies in implementation and integration, not in waiting for new breakthroughs. This process historically takes time for many things: electricity, the internet, smartphones. It will take some time for AI as well. Based on our knowledge working to make this a reality, we can say that it possible to automate around 80-90% of cleaning management in 2024 for leading edge companies like TIDY. Others will likely follow, and improvements will be made where by 2026 we should see about 99% of tasks being able to be automated at the leading edge.

Coming Next

In Part 3 of our series, we'll explore a real-world case study of AI implementation in property management. We'll take a deep dive into how TIDY is putting these principles into practice, examining their progress in solving the challenges of cleaning and maintenance management, and discussing their approach to AI integration from a product perspective. This should give a sense of how the AI revolution will in practice actually roll out, based on our learnings so far.