a16zCan AI Fix Housing and Healthcare Affordability?
CHAPTERS
Elise AI investment and the mission: autonomous buildings to cut major household costs
The hosts introduce a16z’s investment in Elise AI and tee up the central thesis: housing and healthcare are the two biggest household expenses, and AI can remove massive operational waste. Minna frames the company’s ambition as enabling “fully autonomous buildings” and, more broadly, reducing the cost burden these sectors impose on society.
- •a16z announces investment in Elise AI; founders Minna Song and Tony Stoyanov join
- •Housing + healthcare consume ~42% of typical household spending and ~40% of GDP
- •Elise AI’s north star: fully autonomous building operations
- •Core belief: tech can improve experience and lower costs in under-digitized sectors
Why housing and healthcare haven’t been “eaten by software” (and why AI changes the timing)
Minna and Alex discuss why these industries remained expensive while tech-driven categories got cheaper. They argue AI is a paradigm shift because it can finally handle the operational, administrative, and communication burden that legacy software couldn’t.
- •Housing/healthcare prices trend up while tech-touched industries trend down
- •Historically, complex real-world workflows resisted traditional software
- •AI enables automation of admin and communication-heavy work at scale
- •Elise started in 2017; founders went deep in housing and expanded to healthcare
- •Customer outcomes and adoption momentum validate the approach
Housing affordability starts with supply—but utilization and responsiveness matter now
Minna emphasizes supply as the primary driver of affordability, citing a multi-million unit shortfall and insufficient annual building. In the near term, she argues AI can increase utilization of existing stock by fixing broken leasing workflows that leave demand unanswered and units sitting vacant longer than necessary.
- •US shortfall: ~5M housing units; need ~1.8–2.0M new units/year to keep up
- •Recent delivery (~1.5M) implies ~50% increase needed; pipeline expected to shrink
- •Near-term lever: faster conversion of vacancies to occupancy
- •Almost half of rental inquiries go unanswered; “ghosting” creates underutilization
- •Elise data: buildings using its AI showed ~2% higher occupancy vs. market
Why supply is headed the wrong direction: zoning, but also returns and capital flows
The conversation shifts to structural blockers: restrictive zoning and insufficient capital for construction. Minna argues that even with regulatory relaxation, capital must be attracted by improving housing investment returns—something better operations and higher NOI can support.
- •Zoning and regulation contribute to constrained supply
- •Regulatory reform alone is insufficient without capital availability
- •Capital chases returns; housing competes with other asset classes
- •Thesis: “10x better operators” can raise profitability and attract more building capital
- •More supply ultimately is “king” for fixing the housing crisis
YIMBY momentum and real-world examples: what reforms can do
Tony discusses whether full YIMBYism is plausible and points to early proof in Minneapolis. The group explores how quickly supply could respond if building constraints were loosened substantially, arguing the market would build given permission and incentives.
- •Hope for YIMBY progress varies by city; some are moving faster than others
- •Minneapolis example: ending single-family zoning (2019) linked to faster supply growth
- •Reported outcome: rents stayed relatively flat compared to national increases
- •If allowed to build, developers likely will; impacts take years to fully show
- •More competition could spur innovation in housing delivery
Making real estate a better asset: labor, insurance, and controllable operating costs
Minna outlines why real estate operations are inefficient and why returns get pressured—especially by labor costs. She argues AI can counter cost inflation by automating workflows, improving compliance, and optimizing preventative maintenance to reduce downstream expenses.
- •Real estate underinvested in tech; lacks efficiency gains seen in SaaS/internet firms
- •Major headwinds: rising labor costs (especially post-COVID), insurance, supply chain
- •Labor is often the biggest controllable expense for operators
- •AI can reduce manual work, improve compliance, and lower legal/operational friction
- •Preventative maintenance optimization can reduce CapEx surprises and failures
Efficiency without new builds: faster turns, better layouts, and metro-area connectivity
For high-demand cities with limited new supply, Tony highlights ways to improve affordability via utilization and flexibility. These include reducing vacancy time, exploring smaller units and shared amenities, and improving infrastructure so surrounding areas effectively expand supply.
- •Even without new units, utilization can improve (example: SF vacancy ~3.5%)
- •Levers: fill units faster, turn units faster, remove manual leasing friction
- •Design/market ideas: smaller apartments, shared amenities, flexible layouts
- •Infrastructure improvements (e.g., regional transit) expand effective metro supply
- •Acknowledgement: these are partial fixes; building more remains central
From 100 to 200+ units per employee: the path to fully autonomous building ops
Minna describes Elise AI’s goal of enabling portfolios to run core operations with minimal human intervention, leaving primarily physical work and legally required tasks. She shares examples of centralized operating models where AI enables dramatic staff-to-units ratios across many properties.
- •Vision: “fully autonomous buildings” for core operational workflows
- •Most on-site work is administrative/logistical and automatable
- •Physical constraints remain: hands-on maintenance and legal requirements
- •Case examples: Equity Residential (~200 units/employee) and Brookfield centralization
- •Extreme centralization example: specialized role supported by AI across ~10,000 units
What’s automated today: maintenance triage, leasing, self-touring, and documentation
The founders walk through current automation wins across residential operations. They highlight tangible improvements in resident experience, including faster work-order completion and significantly reduced time from listing to lease due to 24/7 touring and instant responses.
- •Maintenance automation: AI triage, prioritization, routing, tracking
- •Work order completion improved from ~4–5 days to under 48 hours in some ops
- •Leasing automation: AI handles repetitive Q&A and high-volume inquiries
- •Touring automation: self-guided tours via smart locks/lockboxes with AI assistance
- •Documentation automation reduces copy/paste work and standardizes processes
Second-order automation: coordinating resources across a multi-building ecosystem
Tony argues the next wave of gains comes from optimizing at the portfolio/ecosystem level rather than per-building. Sharing labor, tools, and parts across properties increases complexity, but enables much larger efficiency improvements—especially for maintenance operations.
- •First-order automation optimizes a single building’s workflows
- •Second-order automation coordinates resources across many buildings
- •Resource pooling: technicians, tools, parts, and scheduling across properties
- •Maintenance expected to see especially large gains from ecosystem planning
- •Higher complexity becomes tractable with AI-driven orchestration
Humans in an AI-run future: community, specialization, and supervising AI systems
Minna explains that as AI absorbs routine communication and logistics, roles shift rather than vanish. Staff move toward relationship-building, conflict resolution, and specialized functions, and eventually toward managing and auditing large fleets of AI agents.
- •Routine/menial tasks decline; resident expectations for frictionless service rise
- •New roles: community engagement, resident experience, conflict resolution
- •Specialization emerges (e.g., renewals/retention specialists)
- •Maintenance remains physical but becomes more efficient and targeted
- •Labor shortage pressure: many technicians are older; efficiency is increasingly necessary
Long-term housing future: robotics, modular construction, longevity, and mobility
The group zooms out to how AGI, robotics, and longevity could reshape housing demand and supply. They connect lower cost of living to family formation and argue robotics/modular methods could reduce construction costs, while AI-enabled operations could increase mobility through shorter, easier leasing cycles.
- •Longevity and wealth effects could increase population and housing demand
- •Cost of living affects birth rates; affordability improvements could change demographics
- •Robotics/modular construction could accelerate building and cut construction costs
- •AI can enable shorter, more flexible leases by reducing turnover labor burden
- •Greater mobility supports jobs, schooling choices, and quality-of-life improvements
Why real estate R&D is so low—and why AI may finally unlock adoption
Minna argues the operational “search space” in housing is huge and full of edge cases, making prior software too brittle and pushing firms to rely on people instead. AI can handle variability, and as automation becomes feasible, real estate could shift from low R&D to significant AI investment.
- •Real estate spends among the least on R&D despite massive spend and complexity
- •Edge cases and building-to-building variability made legacy software ineffective
- •Reliance on people reduced incentives to digitize workflows and collect data
- •Critical operational knowledge often lived “in people’s heads,” not systems
- •AI’s ability to handle complexity makes broad automation and data capture viable
PropTech criticism and the case for adoption: efficiency, competition, and consumer surplus
Responding to critiques that PropTech helps landlords extract more value, Minna and Tony argue technology generally improves service and lowers costs. They claim inefficiency raises barriers to entry and strengthens incumbent pricing power, while mass adoption and competition should pass gains to consumers.
- •Founders reject the idea that keeping processes inefficient protects consumers
- •Analogy: airlines, supermarkets, and other sectors improved with digitization
- •Operational complexity and manual work raise barriers to entry and pricing power
- •Technology tends to create surplus that competitive markets distribute over time
- •Key requirement: mass adoption for competitive pressure to transmit savings
Repairs, maintenance, and operations: scheduling, routing, purchasing, and prevention
Tony and Minna explain how unit turns and maintenance delays drive vacancy and cost, often due to avoidable coordination failures. They see AI as a planning engine that can encode real-world dependencies (what must happen first), reduce lost information, and optimize preventative replacement cycles.
- •Maintenance/turn delays often come from missing info, parts, or poor scheduling
- •AI opportunity: complex planning—routing, sequencing, and technician scheduling
- •Encoding task dependencies and “common sense” constraints improves execution
- •Preventative maintenance: track end-of-life and replace appliances proactively
- •Even small turn-time reductions can unlock billions in value at scale
Expanding into healthcare: similar admin pain, different domain
Elise AI’s move into healthcare is framed as a natural extension of its strengths in administrative workflows. Tony argues the similarities are strongest in intake, repetitive inquiries, phone-based unstructured communication, and scheduling—areas where their voice and ops tech transfers well.
- •Healthcare focus is primarily on administrative (not clinical) workflows
- •Shared dynamics: bloated costs, staffing struggles, regulated environments
- •Common tasks: intake of structured info (names, insurance, preferences)
- •High-volume repetitive questions still handled via phone conversations
- •Housing-built voice and scheduling optimizations translated with few surprises
Why healthcare costs stay high: elastic demand plus admin costs rising fastest
Tony argues both are true: people want more and better healthcare as it improves, but the administrative experience hasn’t improved proportionally. He suggests admin costs have grown faster than clinical value, and AI can finally reduce that burden because it handles unstructured interactions better than prior tools.
- •Healthcare demand is elastic: lower costs could increase consumption
- •Admin experience hasn’t improved much despite rising costs
- •Admin costs have increased faster than clinical-side improvements
- •Prior tech wasn’t good enough for phone-based, unstructured workflows
- •AI can “bend the curve” by automating admin and communication-heavy processes
Where the healthcare platform goes next: from scheduling to billing and longitudinal engagement
Tony outlines a roadmap from the initial interaction through billing and post-appointment communication. The group highlights adherence as a major lever: AI can reinforce care plans, answer follow-up questions, bridge language gaps, and involve family members who weren’t present at appointments.
- •Healthcare is earlier-stage for Elise; large admin surface area remains
- •Potential expansion: intake → scheduling → billing cycle → ongoing patient comms
- •Post-appointment engagement is weak today; AI can extend support beyond the visit
- •Adherence can improve via reminders, education, and Q&A after the appointment
- •AI can address language barriers and include caregivers/family in the loop
Lessons learned and the ultimate vision: tackle the hardest, most underserved—and cut the 42% burden
Minna reflects that she would have started with affordable housing because it has maximal complexity and administrative drag, mirroring their approach to underserved segments in healthcare. They close with an ambition to materially reduce the share of household spend going to housing and healthcare through broad efficiency gains.
- •Retrospective: start with affordable housing due to dense compliance and admin drag
- •Underserved segments offer the largest impact but adopt more slowly
- •Healthcare approach mirrors this: focus on complex, underserved admin problems first
- •Ultimate goal: cost reduction so housing/healthcare aren’t major cost anxieties
- •Target vision: reduce combined share of household spend from ~42% to ~20-something%
