The First AI Governance Framework for Multifamily Housing

Introduction

At OPTECH 2025, RETTC released the first AI Governance Framework¹ for rental housing and technology partners. RealPage was proud to contribute to that effort alongside other housing providers and technology innovators.

The RETTC governance framework gives the multifamily industry a shared vocabulary for how AI should be governed, built around eight principles spanning organizational philosophy, fairness, transparency, privacy, accountability, renter experience, responsible innovation, and third-party due diligence. For an industry that had no common standard before November 2025, that is significant progress.

The principles are the right ones. And the real measure of any AI governance framework is what happens when organizations operationalize those principles and can demonstrate they are working. That is what showing your work means.

Why the RETTC Framework Matters

Establishing Baseline Expectations

The First AI Governance Framework for Multifamily Housing The RETTC framework matters because it establishes baseline expectations for an industry deploying AI at scale into housing decisions. It aligns the industry around key regulatory frameworks including fair housing, fair credit reporting, and accessibility requirements. It puts third-party accountability on the table, encouraging operators to demand transparency from their technology partners regarding data sources, model limitations, and known risks.

That last point is especially important. For years, operators evaluated technology vendors on features, integration, and price. The framework signals that AI governance practices should be part of that evaluation. That is a meaningful shift for multifamily.

Applying AI Governance Principles

From Policy to Practice

Applying AI Governance Principles: From Policy to Practice Every principle in the AI framework raises the same practical question. How do you demonstrate it? Not as a policy statement on a website, but as operational evidence that governance is working. The principles are the foundation. The methodology, measurement, and evidence behind them are what make governance real.

Fairness and Consumer Protection

The framework establishes that technology partners should promote fairness and consumer protection. At RealPage, we had to figure out what AI fairness testing looks like when the artificial intelligence is having a conversation instead of producing a score. A classical screening model outputs a number. You can run a disparate impact analysis against that number across protected classes and measure whether outcomes are disproportionate. The methodology is well understood. A leasing agent AI does not produce a score. It produces a conversation. Fairness testing for conversational AI means evaluating whether the agent responds differently to different types of inquiries, whether tone or guidance varies in ways that could correlate with how a question is phrased, and whether the agent steers some prospects toward touring and others toward the application without a clear reason. We built our evaluation approach around these questions because the framework gives us the principle and the operational work is translating that principle into measurement methodology.

Privacy and Data Integrity

The framework says protect privacy and promote data integrity. For our classical ML models, we monitor for drift in the statistical relationship between inputs and outputs. When explanatory power degrades, we investigate and potentially retrain. For our generative AI agents, the model itself did not change because we did not train it. But a new local ordinance passed, a rental property changed its pet policy, or an evolving fair housing requirement changed how a question should be answered. The drift is in the operating environment, not in the model. Governance has to stay connected to real-world changes, not just model performance metrics, and that requires a fundamentally different monitoring approach than classical ML.

Transparency and Explainability

AI transparency and explainability follow the same pattern. For a classical ML model, explainability means describing the features, the objective function, and why the model produced the output it did. For an AI agent that composes a unique response to every resident interaction, explainability requires tracing a response back through prompt architecture, guardrails, and the foundation model’s reasoning. Operationalizing this principle means building the infrastructure to answer those questions at scale, not just acknowledging that transparency matters.

Accountability and Oversight

Accountability in the framework means establishing processes to review model outcomes and providing escalation channels. In practice, for a generative AI multifamily agent handling thousands of conversations a day, that means defining what an evaluation cadence looks like, deciding when a hallucination is a governance event versus a quality issue, and categorizing failure patterns so they feed back into real improvements. The principle points in the right direction. The methodology is where governance becomes demonstrable.

Building Effective AI Governance

The Need for Layered Controls

None of this works without layered controls. Architecture review before building, risk assessment before deployment, continuous evaluation in production, and regular red teaming. The governance surface must expand alongside the capability surface, especially as the multifamily housing industry moves deeper into agentic AI where agents take actions and coordinate across the full resident lifecycle. A single control point is never sufficient. Defense in depth is how you build governance that holds up under real operational pressure.

Evaluating AI Vendors

Asking for Evidence

Evaluating AI Vendors: Ask for the Evidence The RETTC AI Governance Framework is exactly what our industry needed. It created shared expectations at a moment when property management AI is moving faster than most organizations can govern. RealPage is aligned with these principles and committed to advancing them.

The operators and technology partners who treat these principles as the foundation and invest in the methodology and evidence to back them up will lead the next phase of responsible AI in multifamily.

Ask your AI technology partners to show their work. The answers will tell you a lot.

References

1.  RETTC AI Governance Framework.  Real Estate Technology and Transformation Center, released November 19, 2025, at OPTECH.