RealPage has launched its Lumina AI Suite, a governed intelligence platform that connects property operations, portfolio analytics, institutional intelligence, and open model access. The company also introduced its Five Principles of Trusted AI for Rental Housing, a framework designed to guide how AI reasons, recommends, and remains auditable across the multifamily industry.
The principles require AI to reason from industry-specific data rather than the open internet, enforce one governed definition for every metric, and show the work behind every recommendation. They also mandate that AI operates within existing compliance frameworks with logged actions and that a person remains accountable for all AI-supported decisions. Chief information officer Lance French said the framework gives teams and customers "a consistent basis for evaluating and scaling AI as its reach and authority grow."
Why transparency drives adoption
French identified the principle requiring every recommendation to show its work as the one most likely to accelerate industry adoption. Operators gain confidence when they can see what informed a recommendation, which definitions applied, and what guardrails shaped the output. That visibility makes outputs easier to test, challenge, and improve.
"Evidence becomes meaningful when someone is responsible for interpreting it and acting when the system is wrong," French said. The transparency principle works directly with the requirement for human accountability, creating conditions for effective oversight without removing the need for separate controls around accuracy, fairness, and security.
Addressing operator concerns about risk
Each principle targets a practical concern. Industry grounding provides relevant context so AI does not pull from irrelevant sources. Governed definitions preserve consistent meaning across operations and reporting. Auditability keeps compliance obligations visible rather than buried in model behavior.
French emphasized that the principles function as one system. Their value depends on implementation through architecture review, use-case risk assessment, testing, red teaming, audit trails, monitoring, and escalation. The level of evidence and oversight should rise with the potential consequence of the use case. Fairness testing must also match the capability - evaluating a conversational leasing agent requires examining tone, routing, and escalation across varied interactions, while a scoring model demands different statistical and outcome testing.
The compliance challenge after deployment
The greatest challenge, French said, will be keeping governance synchronized with the full system after deployment. A foundation model may stay unchanged while data sources, prompts, guardrails, property policies, local requirements, integrations, or an agent's action authority shift. Any of those changes can affect how the system performs and which obligations apply.
French also addressed the regulatory outlook, noting that existing fair housing, fair credit reporting, consumer protection, and privacy laws already apply when AI is used in those activities. States are introducing requirements for higher-impact automated decisions. He expects the United States to continue with layered federal, state, and sector-specific expectations, with the most effective standards being risk-based, outcome-focused, and clear about the respective responsibilities of technology providers and housing operators.
What operators get wrong about AI
French pointed to a persistent misconception: that the model itself is the AI system. The outcome a customer experiences is shaped by the complete workflow - source data, governed definitions, retrieval, configuration, identity and permissions, integrations, guardrails, monitoring, fallback paths, and human responsibilities. The same foundation model can support a low-risk internal task and a higher-impact housing workflow with very different governance requirements.
"The material questions concern the job the AI performs, the data and people it touches, the authority it has, and the consequence if it is wrong," French said. Customer conversations have shifted from whether an AI governance policy exists toward the evidence behind it. Operators now ask about success criteria, performance benchmarks, testing protocols, risk classification, and the handoff points between AI and people.
For executives navigating this shift, French advised starting with the actual workflow and a named accountable owner. Define the intended outcome, the data and permissions required, the limits, the human handoffs, and the evidence required before scaling. "Ask every technology partner to show their work, and make sure your own organization can show its work as well," he said. That discipline, he added, gives teams the clarity to move faster while managing risk.
Why this matters for real estate and construction professionals
For multifamily operators and construction firms evaluating AI, the RealPage framework provides a concrete checklist for vendor assessment. The five principles translate into specific questions to ask any technology partner: Does the system reason from industry-specific data? Are definitions governed and consistent? Can every recommendation be traced to its inputs and guardrails? Are actions logged and auditable within existing compliance structures? And who is accountable when the system is wrong? Professionals who build their own AI adoption around these questions will be better positioned to scale tools without creating compliance exposure. Those looking to build internal capability can explore AI for Real Estate Courses or, for leadership roles, AI IT Strategy Training to develop the governance discipline French describes.
Your membership also unlocks:
