Affordable Housing Investors Council
Create Account

AI in Affordable Housing: Key Questions for Equity Investors

Anne Hollander opened the session by defining AI as a pattern-finding tool that simulates aspects of human intelligence—learning, decision-making, and creativity—by identifying probabilities across vast datasets. In the context of affordable housing, AI helps organizations transition from a data-saturated environment into an "insight age," where automation and pattern recognition generate more actionable intelligence. (View the slides, watch the full video)
 
Anne emphasized three core outcomes where AI can create value:

The Transformation Curve: Five Stages of AI Maturity

Hollander mapped out a transformation curve for AI adoption:

  1. Readiness & Education: Organizations set AI policies, educate staff, and assess use cases.

  2. Early Task Automation: Routine work—like naming and storing documents—is automated, saving time and reducing human error.

  3. AI Assistance: Tools like ChatGPT and Microsoft Copilot are deployed to support email management, research, and productivity.

  4. Deeper Automation: More sophisticated uses emerge, such as predictive analytics and integrated reporting.

  5. Strategic Differentiation: Organizations use AI to build real-time ecosystems and proprietary insights, driving sustained competitive advantage.

Nine Questions Every Investor Should Ask Operators and Managers

Anne offered a powerful due diligence framework structured around nine key questions, helping investors distinguish between aspirational AI use and scalable, accountable implementation.
 
1. What measurable lift are you seeing or targeting from AI tools?

Green flags: Specific KPIs and ROI within 3–6 months; examples like improved collections or reduced utility costs.
Red flags: Vague pilots, buzzwords, or statements like “we’ll see lift once the vendor is implemented.”

2. How does your AI strategy strengthen financial resilience?

Green flags: Use cases linked to income, costs, reserves, or CapEx with pre- and post-AI comparisons.
Red flags: No baseline metrics, generalized claims of "cost savings."

3. How are AI models validated and monitored?

Green flags: Governance frameworks, performance monitoring, accountability pathways.
Red flags: No formal validation process or deferring responsibility to vendors.

4. What AI governance framework do you have in place?

Green flags: Written policies, periodic audits, legal review, and clear documentation of high- and low-risk use cases.
Red flags: IT “keeps an eye on it,” or there’s no distinction between AI and software governance.

5. What is your incident response plan for AI-related errors?

Green flags: Documented incidents, root cause analyses, and escalation plans.
Red flags: “We’ve never had an issue” or "We'll shut it off if something goes wrong."

6. How is sensitive or confidential data protected?

Green flags: Encryption, cyber insurance, data retention settings, and vendor agreements with oversight.
Red flags: “We don’t put sensitive data into AI,” or assuming ChatGPT deletes data (it doesn’t entirely).

7. What tasks are automated or augmented by AI, and what KPIs are tracked?

Green flags: Before-and-after dashboards, reduced response times, and productivity metrics.
Red flags: Unclear task changes, no data to support performance gains, or discussion of AI replacing staff.

8. How is your team trained to use AI responsibly and effectively?

Green flags: Dedicated training budgets (~$500 per FTE), adoption targets, usage tracking, and incentives tied to KPIs.
Red flags: "Learn on your own" approaches, no tracking, or shelfware tech that isn’t being used.

9. How are AI outputs integrated into workflows without creating data silos?

Green flags: APIs or orchestration platforms that connect systems of record and systems of engagement.
Red flags: Manual Excel imports/exports and no clear integration strategy.

Change Management and Organizational Readiness

Beyond tech deployment, Hollander stressed the importance of organizational change management. She noted that AI-related changes typically touch four areas: technology, processes, people, and culture. Failing to align these dimensions can undermine even the best AI tools.

For example, small automations—like naming documents or triaging email—can help introduce staff to AI without overwhelming them. Starting with these "headache tasks" reduces fear of job displacement and builds internal champions.

Affordable Housing's Unique Opportunity

While the real estate sector is traditionally risk-averse and slow to adopt technology, Hollander believes affordable housing is uniquely positioned to lead on AI. The sector’s deep operational complexity, resource constraints, and mission orientation create high demand for tools that improve efficiency and service quality. Compared to other real estate subsectors, affordable housing leaders show greater openness to adopting AI in a purposeful, incremental way.

What Excites and Worries AI Experts

When asked what excites her most, Hollander pointed to the potential for “data rails” in affordable housing—shared infrastructure enabling seamless information flow across operators, investors, and regulators. This could transform underwriting, compliance, and decision-making at scale.

Her biggest concern? The low barrier to entry for AI vendors. Tools built without an understanding of affordable housing—and without adequate security, governance, or transparency—pose serious risks. She urged investors to evaluate vendors carefully and avoid over-reliance on unproven platforms. 

Looking Ahead

AI is not a magic bullet—but in affordable housing, it is a tool with real potential to amplify mission impact, improve outcomes, and modernize operations. As Anne Hollander emphasized throughout the session, it’s not about chasing the latest tech. It’s about building a durable strategic advantage.