Enterprises Have The Claude License. Now What?

RealComm San Diego is where commercial real estate's technology leadership takes stock. Not of vendor demos or product roadmaps, but of what is actually working inside the enterprise — and what still isn't. At the June 2026 conference, the room was predominantly CIOs managing office, retail, and industrial portfolios. And the conversation, once it got past the keynote optimism, kept landing on the same three pressure points: AI adoption, AI governance, and token visibility.

These are not abstract concerns. They are the operational realities sitting on the desk of every CRE technology leader right now, beneath the mandate to "accelerate AI" that came down from the executive suite sometime in the last eighteen months.

Three New Problems on Every CIO's Plate

AI Adoption

The license exists. The executive mandate exists. What doesn't exist, in most organizations, is a structured path from both of those things to a team that actually operates differently. Adoption in most CRE firms is shallow — a handful of power users, a lot of people who opened the tool once, and a long tail of licensed seats generating no value. The CIO is accountable for that gap. They didn't create it, but they own it.

AI Governance

Who is allowed to use AI for what? What data can enter a prompt? What outputs can be acted upon without human review? What happens when a junior analyst feeds sensitive deal data into a model? Most CRE firms do not have clear answers to these questions yet, and the absence of policy creates its own risk — either the tool gets over-restricted and adoption stalls, or it gets under-restricted and something goes wrong. CIOs at RealComm were explicit: governance frameworks are the thing they most need help building, and they are building them largely from scratch.

Token Visibility

For CIOs managing enterprise AI spend, token consumption is the new software license audit. Usage is distributed, often invisible at the departmental level, and difficult to attribute to specific workflows or outcomes. Whether the firm is getting ROI from its AI investment is almost impossible to answer cleanly without visibility into how tokens are being spent, by whom, and on what. This was a more technical conversation than it sounds — and it was happening at every table.

Why CIOs Are Advocating for Claude Enterprise

Against this backdrop, the Claude Enterprise license has earned genuine advocates in the CRE CIO community — not because it is the loudest brand in the room, but because it solves a management problem that others don't address as cleanly.

The core appeal is simplicity. Claude Enterprise gives CIOs a single surface for managing users, MCP integrations, and token consumption across the organization. In an environment where governance and visibility are top priorities, that matters more than marginal model performance differences. You can see who is using what, set policies at the organization level, and integrate internal tools through MCPs without rebuilding your stack. For a CIO already managing a sprawling technology estate, that operational clarity is genuinely valuable.

The concern — and it surfaced openly at RealComm — is vendor concentration. If Anthropic becomes the default enterprise AI infrastructure for a significant portion of commercial real estate, pricing power shifts. CIOs who have been through enterprise software vendor cycles before recognize the pattern. The tool becomes essential, switching costs compound, and the next renewal conversation happens on different terms.

This is why most CRE CIOs are pursuing a deliberate dual-track strategy: Claude Enterprise as the primary operational platform, Microsoft 365 Copilot as the hedge. Not because Copilot is preferred on its merits, but because maintaining a credible alternative preserves negotiating leverage and reduces single-vendor exposure. It is less a technology decision than a procurement philosophy.

The SaaS Question Has Been Settled

At the CIO level in commercial real estate, the verdict on purpose-built AI SaaS is in. The language is precise — less frustration, more strategic clarity — but the conclusion is firm: point solutions are not the answer.

The reasoning is structural. Every AI SaaS product evaluated adds three things CIOs don't want: another vendor relationship to manage, another integration surface to maintain, and another data governance exposure to audit. The ROI case for a point solution has to clear a very high bar when the organization already has a horizontal AI platform it is trying to drive adoption into.

What CIOs said they prefer is the incumbent-first model: existing platforms — the property management system, the lease management platform, the ERP — with AI capability layered in by the vendor. That keeps the data model clean, the integration surface familiar, and the governance perimeter manageable. Where the incumbent falls short, the gap gets filled by Claude operating as an augmentation layer — something like Claude Cowork or a custom internal deployment — rather than a new system with its own login, its own data model, and its own support contract.

This is a more sophisticated position than "we don't want new software." It is a coherent architecture for how AI gets embedded without exploding the technology estate.

The Build-vs-Buy Reality

The question that followed naturally from every governance conversation was whether firms should build Claude skills in-house. The honest answer from CIOs: probably yes, eventually — but not now, and not with current resources.

Engineering teams in commercial real estate are not large. They run lean against a backlog of infrastructure work, integration debt, and the ongoing demands of running mission-critical systems for a portfolio. The capacity to build and maintain custom AI workflows — prompt engineering, MCP configuration, agentic workflow design, ongoing iteration as Claude's capabilities evolve — simply does not exist at most firms without either hiring specifically for it or pulling engineers off work that is already understaffed.

The Anthropic joint venture with Blackstone, Goldman Sachs, and Hellman & Friedman, announced in May 2026, named this problem explicitly. Blackstone President Jon Gray framed it as "one of the most significant bottlenecks to enterprise AI adoption" — the scarcity of engineers who can implement frontier AI systems at speed. The $1.5 billion venture is designed to embed Anthropic's engineers and models directly into the operations of mid-size businesses, and it carries a signal worth reading carefully: if Blackstone concluded that the answer to the AI implementation gap was bringing in domain-plus-Claude specialists rather than building the capability entirely in-house, that tells you something about the true cost of the DIY path. (Source: Blackstone press release, May 2026)

Blackstone's own release noted that Claude's capabilities change monthly or even weekly — making AI deployment fundamentally different from traditional enterprise software rollout. That observation matters for CIOs considering the in-house build path. What the internal team builds to today may be superseded before it reaches the full user base.

The implication is not that in-house capability is irrelevant. It is that the first phase — building the initial workflow library, establishing governance templates, training the team — is better done with outside specialists who already carry both the domain knowledge and the Claude fluency, and who can transfer both to the internal team in a structured way.

The Gap That Remains: Training at the Human Level

Large-scale ventures like the Blackstone-Anthropic partnership solve the enterprise engineering problem. They do not solve the practitioner training problem.

CIOs at RealComm were clear-eyed about this distinction. Getting Claude deployed in the organization is an engineering and governance challenge — addressable with the right infrastructure partner. Getting the asset manager, the property manager, the leasing director, and the in-house lawyer to actually use it well is a different challenge entirely. It is a skills challenge. And it is not solved by an API integration or an MCP configuration.

The pattern is consistent across every enterprise AI deployment: the platform gets stood up, a few technically curious people adopt it, and everyone else continues working the way they always have. The productivity gains that justified the license never materialize at scale because the human layer — the actual practitioners doing the work — was never given the structured time and domain-specific guidance to build new workflows into their practice.

This is not a failure of will. It is a failure of format. Generic AI training programs, vendor onboarding sessions, and internal lunch-and-learns do not produce practitioners who can build and run agentic workflows in their specific domain. They produce people who have watched a demo.

What produces practitioners is a structured, role-specific working session — not a presentation about AI, but a session where participants actually build the workflow for their job, with their data, in the tool their firm already has, guided by someone who understands both the technology and the professional context they operate in.

My intensive program is built for exactly this moment in a CRE organization's AI journey: after the license is purchased, after governance policy is drafted, after engineering has stood up the platform — and before any of that investment translates into daily practitioner behavior.

For CIOs, the value proposition is direct: this is the activation layer that makes the license spend defensible. The platform investment is already made. The session converts that investment into measurable practitioner behavior change — without adding to the engineering backlog, the vendor roster, or the governance surface.

The Takeaway from San Diego

RealComm 2026 made clear that commercial real estate's CIO community has moved past the question of whether to adopt AI. That decision is made. The questions now are operational: how to govern it, how to see it clearly, how to build skills at scale without overloading the engineering team, and how to get practitioners — not just engineers — working differently.

The Blackstone-Anthropic partnership validated the domain specialist model for enterprise AI implementation. The CIO consensus validated Claude Enterprise for governance and token management. What neither addresses is the last mile: the individual practitioner, in their specific role, building the habits and workflows that make the license worth what the firm paid for it.

That gap is real, it is widely acknowledged, and it is closeable. The question is whether your organization closes it in 2026 — or spends another year watching the license sit underused while the firms that did close it pull ahead.