MAAP for Senior Housing & Care: Models, Agents, Applications — and the Operating Record

Peter Gassner, CEO of Veeva Systems, recently framed the next era of enterprise technology as the MAAP architecture — Models, Agents, and Applications. The framing is correct, and it is overdue. Enterprise technology is no longer infrastructure that humans use to do work. It is increasingly the work itself, with agents reasoning over models and executing against applications. Greg Meyers, CDTO of Bristol Myers Squibb, sharpened the implication: the application layer itself has to be rethought from first principles, because today's UIs were built to compensate for human variance, and agents do not need that scaffolding.

Both observations land squarely on what SeniorCRE has been building for senior housing & care. We agree with the architecture. We want to push it further into our industry, because the implications for operators, owners, lenders, and investors are not theoretical — they are governance decisions that get made in the next twenty-four months.

I. What MAAP Actually Says

In Gassner's framing, the three layers each do something different:

• Models — frontier systems (Anthropic, OpenAI, Gemini, and others) that reason, generate, and interpret. They are increasingly capable, increasingly substitutable, and increasingly commoditized at the foundation level.
• Agents — a new class of knowledge worker. Agents use models for reasoning and applications for execution. They take on tasks that previously required humans, and they need to be governed like a workforce, not configured like a feature.
• Applications — the systems of record. They codify the business rules of an industry and the operating model of a specific company. They provide the structure and consistency that make the business legible — to humans, to agents, and to capital.

The strategic claim inside MAAP is that applications are not being disintermediated by AI. They are being elevated. As models commoditize and agents proliferate, the system that holds the trusted state of the business — the operating record — becomes the most valuable layer in the stack, not the least.

II. Why Senior Housing & Care Is the Canonical MAAP Industry

Veeva built the dominant system of record in life sciences because life sciences is a regulated, multi-stakeholder, capital-intensive industry where one trusted operator-controlled operating record is worth more than ten well-designed point tools. Senior housing & care meets every one of those conditions, and arguably meets them more acutely.

• Regulated. CMS, state survey agencies, HIPAA, payer audits, MDS and Section GG documentation, SNF QRP validation, RIDEA structures, and lender covenants all demand a defensible operational record.
• Multi-stakeholder. A single resident touches operators, clinicians, families, payers, REIT owners, lenders, and increasingly, AI systems. They all need to reason against the same truth.
• Capital-intensive. The asset only performs if the operation performs. Capital partners are no longer satisfied with month-end packets. They want continuous operational visibility.
• Fragmented. Most operators run between seven and twenty discrete systems. There is no system of record today — only a constellation of departmental tools that an army of humans reconciles by hand.

That last condition is the opening. In the MAAP era, fragmentation is not just operationally expensive — it is agent-hostile. An agent cannot reason reliably against twelve disconnected systems with twelve different definitions of "resident," "shift," "occupancy," or "revenue." Every additional system is another translation layer, another failure mode, another reason the agent's output cannot be trusted by a clinician, a regional VP, or an investor.

III. The Operating Record Is the Moat

SeniorCRE is built around a single architectural commitment: one operator-controlled operating record across the entire stack. Six entities — Resident, Care Plan, Ledger, Shift, Property/Unit, Entity — are defined once, governed once, and reasoned against once. The per-community instance of that model is the operating record.

In MAAP terms, the operating record is the application layer for senior housing & care. It is what agents reason against. It is what models are grounded in when they answer a clinician's question, draft a survey response, or surface an occupancy risk. It is what capital reads when it underwrites a deal or monitors a covenant. It is the trusted execution layer.

SeniorCRE is not valuable because it has AI. SeniorCRE is valuable because AI makes trusted operating data exponentially more important.

Models will continue to improve and continue to commoditize. Agents will multiply across every workflow in the building. What does not commoditize is the operator-controlled operating record underneath them. The community, the portfolio, and the capital stack will reason against whichever record is canonical. That record is the asset.

IV. Dual-Mode Applications, Built for Agents as First-Class Consumers

Greg Meyers's point deserves to be repeated in our industry: today's senior care UIs exist because frontline staff bring wildly varying levels of context, training, and attention to software. Buttons, dropdowns, modal stacks, and 200-field assessments are scaffolding to bridge that gap. Agents do not need any of it. To an agent, the "good UX" of a legacy EHR is often pure friction.

We agree, and we are building accordingly. SeniorCRE applications are designed to be dual-mode from the schema up:

The result is the same business rule expressed twice — once for the clinician on the floor, once for the agent reasoning on behalf of the operator, the REIT, or the lender. The schema is the contract. Whichever consumer shows up, the answer is consistent.

V. Agents as a New Workforce

The most underweighted line in Gassner's post is the workforce one. Agents are a new type of knowledge worker. That is not a metaphor. In senior housing & care, where labor is 55–65 percent of operating expense and where workforce variance is the largest single driver of margin volatility, the arrival of a digital labor pool that can take on documentation, scheduling, billing, compliance monitoring, and family communications is an operating-leverage event, not a feature release.

But it only works if the digital workers can be governed like workers:

• They need a defined scope of authority — what they can read, what they can write, what they must escalate.
• They need a defined accountability trail — every action attributable, reviewable, and reversible.
• They need a defined operating model — pilot, prove, scale, with humans in the loop until the outcome is contractable.
• They need a defined system of record to act inside — not a patchwork of integrations stitched together at runtime.

SeniorCRE's posture on AI has always been governance first → intelligence second → execution last. The MAAP framing makes the reason explicit. You cannot deploy a workforce of agents into a fragmented stack and call it transformation. You can only deploy it into an operating record that the agents — and the humans who supervise them — can trust.

VI. What This Means for CEOs and Capital Allocators in Senior Housing & Care

Gassner is right that this is no longer a technical decision. The choice of operating record is a choice about the company's operating model, its workforce, and its capital profile for the next decade. In senior housing & care specifically, three things follow:

1. The control plane is the claim. Frontier models will keep improving and keep getting cheaper. The agents will keep multiplying. The durable question is who declares what an agent may read, act on, escalate, and cite.
2. Fragmentation gets repriced. Operators running on seven to twenty disconnected systems are not just inefficient. They need explicit authority rules before agent-mediated workflows can govern decisions. As capital partners begin to underwrite operations through governed evidence, fragmentation will show up in the cap rate.
3. Governance moves up the stack. Boards used to ask about ERP. They will start asking about the operating record, the agent inventory, and the authority policy that controls both. The CEO is now responsible for digital labor policy, not just human workforce policy.

VII. The SeniorCRE Position

SeniorCRE is operator-controlled operating infrastructure for senior housing & care: definitions, source authority, reconciliation, lineage, and agent permissions above retained systems. We are building the application layer of MAAP for this industry with human-facing workflows and agent permissions governed by the same operator authority model.

The race in our industry is not which frontier model wins. It is which authority model operators accept. Operators, REITs, and lenders who govern definitions, source authority, reconciliation, and lineage early will be better prepared for operations and capital review. Those who wait will spend it explaining gaps.

MAAP is the right map. In senior housing & care, the operating record is the territory.