Uncovering Runaway Rentals: How AI Can Help Syracuse Enforce Housing Standards

Syracuse’s housing situation is under the spotlight thanks to last week’s City Auditor’s report, “Runaway Rentals,” focusing on the city’s often overlooked Rental Registry. This registry, meant to ensure safe and code-compliant housing, is a database of all one- and two-family, non-owner-occupied rental properties. Landlords are required to apply for a Rental Registry certificate, but according to the report, compliance is shockingly low. Auditor Alexander Marion puts it starkly: “Our analysis believes compliance may potentially be as low as 25% — just one in four rental properties.” This isn’t just a paperwork issue; it’s a matter of resident safety and community well-being.

The Auditor’s report highlights a crucial challenge: data silos. The report identifies a real problem in the way information is managed: the City struggles to accurately track which properties should be registered because of “messy” address standardization. Think of it: Is City Hall located at 233 East Washington St., 233 Washington St E, or 233 E Washington Street? These small differences make matching property information across the City’s different datasets a real headache and ultimately leads to a lack of accountability.

Of course, this piques my interest. How can we make better use of existing data? Can AI bridge these data gaps? I decided to experiment with Google Gemini, and the results were more than just encouraging. They were exciting.

Building a Smarter System: The Syracuse Address Checker

My goal was to create a system that could intelligently connect disparate datasets and ultimately help improve the City’s enforcement of its rental standards. Here’s what I built in a couple of hours:

• Data Aggregation: I pulled data from Syracuse’s Open Data portal, combining the Rental Registry, code violation data, and other relevant sources. You can check out the code here: https://github.com/samedelstein/syracuse_address_checker/syr_open_data.py
• AI-Powered Address Matching: Using Google Gemini, the system now matches an inputted address across different datasets, despite variations in formatting.
• Code Violation Summarization: Gemini not only matches addresses but also provides a concise summary of any outstanding code violations. This is key to understanding the state of a property, even when violations are listed in complicated terms.
• Future-Proofing: I designed the system so it can be easily integrated into existing city processes, such as permit applications and service requests. This ensures that future interactions with the city can trigger a check for rental registry compliance.

The images above showcase the power of the “Syracuse Address Checker” (Code is available on GitHub here: https://github.com/samedelstein/syracuse_address_checker). In these examples, Gemini was able to recognize 1315 Cumberland Ave as a match on the Rental Registry, but no code violations. It also correctly determined that no properties were listed for another address. It recognized “1019 James St Syracuse NY” as “1019 James St” in the code violations database but also that it was not in the rental registry. It summarized the 20 code violations into a paragraph. Most impressively it was able to match 130 Bryant Ave to data related to 130–32 Bryant Ave and summarize the two code violations.

Why this Matters: Beyond the Basics

The report emphasizes that the Rental Registry, “has the potential to be a significant driver of revenue for City housing operations,” but that budgeted projections often fall flat. This work demonstrates that using data and technology, paired with process, could make the program more effective, both in ensuring safe housing and generating expected revenue. This project isn’t just a technical exercise. It is similar to some of the work we did in the API office when I was there, prioritizing neighborhoods for Code Enforcement Officers to visit: TOP Pilot for Community Code Enforcement — Innovate Syracuse

This demonstration is could be a way to:

• Increase Compliance: By identifying non-compliant properties more effectively, we can push more landlords to register their properties and meet code standards.
• Improve Tenant Safety: By highlighting outstanding violations, we can help ensure that properties are safe and habitable for residents. As Auditor Marion says, it’s about making “sure every Syracuse tenant has a safe, healthy home.”
• Streamline City Processes: By linking disparate data, we can streamline city operations and avoid wasting time and resources on manual checks.

Future Use Cases: Expanding the Possibilities

This is just the beginning. Here are some potential avenues for expansion:

• Predictive Modeling: Using machine learning, we can identify properties that should be on the registry, but currently are not. This can lead to more targeted enforcement efforts.
• Proactive Compliance: When landlords apply for permits or ask for services, we could use this system to ensure they are in compliance with the rental registry.
• Enhanced Inspections: Empower code enforcement officials with tools that leverage the system, allowing them to spend more time on inspections and less on paperwork.
• Citizen Engagement: We could even leverage technology to allow citizens to report potential violations and connect photos and data that create work orders.
• Agent Support: Agents could use the tools to conduct more research and tie data together that they could not see before.
• Increased Revenue: The data shows a potential of 15,000 properties that could be on the registry, generating revenue for housing operations. As the report stated, “Actual Rental Registry Revenue Misses Adopted Revenues By Hundreds of Thousands of Dollars Each Year.”

The “Runaway Rentals” report has clearly identified the need for better housing enforcement. By bringing together AI and open data shows that a more effective, responsive, and accountable system can be built to protect residents and improve Syracuse’s communities.