How Artificial Intelligence Could Strengthen Affordable Housing Oversight and Protect Taxpayer Investments
San José has spent years positioning itself as one of the nation’s leaders in government artificial intelligence.
City leaders have promoted GovAI as a tool that can modernize government, improve public services, and help employees work more efficiently. The initiative has generated national attention and established San José as a city willing to embrace emerging technology rather than simply react to it.
That is why City Auditor Joe Rois’ recent audit of the City’s affordable housing portfolio may be one of the most important tests GovAI has faced so far.
Not because the audit is about artificial intelligence.
But because it identifies exactly the kind of operational challenge artificial intelligence was created to solve.
The Housing Department oversees approximately 220 affordable housing developments containing nearly 25,000 affordable housing units and a permanent loan portfolio valued at approximately $745.6 million. These properties represent one of the City’s largest public investments, affecting thousands of residents and hundreds of millions of dollars in taxpayer-supported financing.
These numbers help explain why the audit should not be viewed as a criticism of Housing Department staff. The department is responsible for overseeing 220 properties, nearly 25,000 affordable housing units, and a loan portfolio approaching three-quarters of a billion dollars. Yet according to the audit, the City has only one building inspector assigned to monitor these developments—and while the team has a goal to inspect 105 properties annually, staff estimates a single inspector can realistically only handle 80. The challenge is not a lack of commitment. The challenge is scale.
The audit found monitoring challenges, incomplete reviews, inspection backlogs, software limitations, and gaps in compliance tracking.
The findings raise an obvious question:
Where is GovAI?
The Problem Is Not a Lack of Data
One of the most important conclusions from the audit is that the Housing Department does not appear to suffer from a lack of information.
It suffers from a lack of capacity to process information at scale.
Housing staff are responsible for reviewing rent rolls, tenant income certifications, financial statements, operating budgets, reserve account information, inspection reports, loan agreements, promissory notes, affordability restrictions, amendments, and compliance records across hundreds of properties.
No reasonable person reading the audit could conclude that staff are not working hard enough.
The challenge is volume.
Government increasingly generates more information than employees can realistically review manually.
That is precisely where artificial intelligence can create value.
Modern document intelligence platforms can review thousands of records, identify missing documentation, compare rents against affordability requirements, flag unusual patterns, summarize contracts, and generate exception reports for human review.
The goal is not to replace employees.
The goal is to help employees focus on the files that actually require attention.
The Audit Reveals a Significant Financial Risk
One finding should immediately attract the attention of taxpayers.
The auditor identified a property containing 27 over-income households that were not being charged the higher rents permitted under program rules. According to the audit, the project may have been foregoing approximately $114,000 in annual rental revenue.
At the same time, the development repaid only $1,694 toward a $7.6 million City loan during 2025.
That is not simply a compliance issue.
It is a financial oversight issue.
When affordable housing developments fail to maximize allowable revenue, fewer dollars become available for property maintenance, debt repayment, and future affordable housing investments.
An AI-powered monitoring system could automatically compare tenant income certifications against rent schedules, identify properties charging below permitted levels, and flag anomalies long before they appear in an auditor’s report.
The longer these issues remain undetected, the larger the financial consequences become.
Affordable Housing Compliance Is Becoming More Complex
Housing Director Erik Soliván’s response to the audit highlights another reality that deserves attention.
Affordable housing compliance rarely operates under a single set of rules.
Many developments combine federal HOME funds, Low-Income Housing Tax Credits, tax-exempt bond financing, state housing programs, and local funding sources. Each funding source introduces its own reporting requirements, affordability restrictions, income calculations, and compliance obligations.
The result is a complex web of overlapping regulations that can overwhelm even experienced analysts.
Artificial intelligence is particularly well-suited for this type of structured analysis.
Rather than requiring staff to manually compare multiple regulatory frameworks, AI systems can evaluate layered requirements, identify conflicts, and highlight the most restrictive standards that apply to a particular property.
In many cases, the challenge is not understanding the rules.
It is keeping track of all of them simultaneously.
Before AI, Housing Must Solve Its Data Problem
The audit also reveals a less visible but equally important challenge.
Housing records are currently dispersed across multiple systems, including the department’s primary software platform, internal databases, and employee email accounts.
That should concern anyone responsible for public accountability.
When information is fragmented across multiple locations, institutional knowledge becomes difficult to preserve, records become harder to locate, and oversight becomes more challenging.
If a key employee leaves, important compliance information can become significantly harder to access.
Before San José can fully realize the promise of AI-powered housing oversight, it must first ensure that its records are organized, searchable, and accessible.
The City does not need another flashy technology demonstration.
It needs a centralized knowledge management system capable of bringing together decades of loan documents, regulatory agreements, compliance reports, amendments, inspections, and financial records into a single searchable platform.
Only then can advanced analytics and AI tools deliver their full value.
The Auditor’s Risk-Rating Recommendation May Be the Most Important Finding
Among all of the auditor’s recommendations, one stands above the rest.
Rois recommends implementing a formal risk-rating framework for affordable housing properties.
That recommendation should immediately attract the attention of GovAI.
Risk assessment is one of the most mature and successful applications of artificial intelligence in both the public and private sectors.
A risk model could evaluate reserve balances, vacancy trends, financial performance, inspection findings, repayment history, compliance violations, and property conditions to identify developments that require additional oversight.
Instead of treating all 220 properties equally, Housing staff could focus their limited resources on developments presenting the greatest risk.
This is not a futuristic concept.
Banks do it.
Insurance companies do it.
Asset managers do it.
Regulators do it.
The auditor has effectively provided a roadmap for how artificial intelligence could strengthen affordable housing oversight in San José.
GovAI Needs to Move Beyond Demonstrations
San José deserves credit for embracing innovation.
But innovation should ultimately be measured by results.
The Housing Department oversees 220 affordable housing properties, nearly 25,000 affordable units, and a $745.6 million loan portfolio. The auditor identified inspection backlogs, incomplete reviews, fragmented data systems, compliance monitoring challenges, and opportunities for stronger risk management.
In many ways, the audit describes the exact environment where artificial intelligence should be delivering measurable public value.
If GovAI is already working with the Housing Department on document analysis, compliance monitoring, risk scoring, or portfolio management, City Hall should explain what is being developed and when residents can expect results.
If GovAI is not involved, taxpayers should reasonably ask why one of the City’s flagship technology initiatives is absent from one of the clearest use cases identified by the City Auditor.
The goal should not be deploying AI because it is fashionable.
The goal should be deploying AI where it improves accountability, strengthens oversight, protects public investments, and helps public employees perform their jobs more effectively.
The Community and Economic Development Committee will soon review the auditor’s recommendations and the Housing Department’s response.
Committee members should ask a simple question:
If GovAI was created to solve real government problems, why isn’t it being deployed to help manage a $745 million affordable housing portfolio?
Joe Rois set out to evaluate affordable housing oversight.
Instead, he may have inadvertently identified the most compelling GovAI use case City Hall has seen so far.