Episode 19 of Multipliers: Three Things CRE Firms Getting Ahead Are Actually Doing
Most CRE firms have adopted an AI harness like Claude, ChatGPT, or Copilot, but very few know what to do next.
That’s an accurate description of where the industry is right now. The CEO Spencer spoke with the day before this recording runs a major investment management firm. He has a CTO. He has a plan. He has already committed to an AI harness and is getting varying degrees of value from it. And when Spencer got on the phone with him, the thing he said was: I’m not sure what to do next.
Spencer has heard this enough times now, across hundreds of conversations and engagements with commercial real estate firms, to know that it is a framework problem, not a knowledge problem. The firms that are getting ahead have not found some secret technology. They have done three specific things that most firms have not done yet. That is what this episode is about.
In this episode of the Multipliers podcast, Spencer Burton and Sam Carlson cover the three actions Spencer sees in every CRE organization that is actually getting ahead, why build or buy is the wrong question, why practical AI governance matters more than most firms realize, and what it looks like when one person with the right tools rebuilds in nine days what two engineers built in eighteen months. For CRE professionals and firms looking to implement AI in commercial real estate, this episode lays out the framework.
- You might also enjoy: The prior episode on JEV and why system one models may be a bigger deal for CRE than LLMs: Episode 18 of Multipliers: JEV Is a Bigger Deal for CRE Than LLMs
- Related: Build the technical foundation this episode references: The A.CRE Accelerator
Episode 19 of Multipliers: Three Things CRE Firms Getting Ahead Are Actually Doing
Michael Belasco is off this week. Spencer and Sam open with a detour into Disney World, Walt Disney’s flywheel, and Sam’s thirty-three days at Disneyland last year before turning to the headline story: Spencer just emerged from nine days in a dark interior cruise ship cabin, working twelve-hour stretches to rebuild a major A.CRE digital asset while his wife and daughter enjoyed the ship in a separate room. The same digital asset had previously taken two engineers, eighteen months to build. Spencer rebuilt it, better than before, in nine days. He also leveled up CRE Agents during the same sprint.
The conversation then turns to a call Spencer had with a CEO of a real estate investment management firm the day before recording, which becomes the framework for the episode’s core argument. Spencer runs a company that gives AI the agent skills and real estate data it needs and teaches at UNC Kenan-Flagler. Sam runs UpHex and has been to Disneyland thirty-three times in one year.
The episode connects back to Episode 18, where Spencer and Sam covered JEV and the cost compression happening at the intelligence layer. This one zooms out to the organizational question: given everything the technology can now do, what do firms actually need to do to take advantage of it?
Listen on: Apple Podcasts | Spotify
Why This Episode, Why Now
The reskilling and retooling wave Spencer describes in this episode is happening now, and most firms are in the middle of it without a clear map. Spencer has said this across multiple episodes, but the conversation with the CEO sharpened it in a specific way: here is a sophisticated firm with real resources, a dedicated tech team, and an active AI harness deployment, and the person running it cannot articulate what to do next.
Spencer’s framework for that conversation is built on hundreds of similar ones over the past several years. He has been in the trenches since 2020 at Stablewood, building AI-powered systems for real estate operations. He has trained thousands of CRE professionals on the AI front. He has done engagements with hundreds of firms. The three things he identifies are the observable differences between the firms that are pulling ahead and the ones that are still figuring out how to get value from their Claude subscription.
The cruise ship story matters as context for the same reason. Spencer rebuilt a major digital asset in nine days that previously required two engineers and eighteen months. It also shows that the tools available right now change what a single motivated person with domain knowledge can accomplish. The same logic applies at the firm level: the organizations that figure out how to channel that capability systematically will be well ahead of their peers in three to five years.
Episode Highlights
Here are the themes that stood out.
1. Going Into the Cave: Nine Days, One Person, Better Than Eighteen Months With Two Engineers
Spencer’s cruise ship story is worth understanding in full because it illustrates what is actually possible right now when a domain expert with AI fluency commits uninterrupted time to a build.
Spencer described the role he plays at A.CRE as straddling CEO and CTO. Now that practitioners who understand their domain can build things that previously required dedicated engineering teams, the distinction between those two roles has blurred. Spencer’s first computer interaction was from a command line. He has managed the server and digital assets of A.CRE for twelve to thirteen years as a hobby alongside running the business. With AI, that background has become a real advantage.
The build he needed to do had two parts: a major A.CRE digital asset that needed rebuilding, and a meaningful upgrade to CRE Agents. The last time the A.CRE digital asset was rebuilt, two engineers spent eighteen months on it. Spencer spent nine days on a cruise ship in an interior room, working twelve-hour stretches, coming out only for food, the gym, and an hour or two of dancing with his wife at night. The result was better than the previous version and took a fraction of the time.
Spencer’s wife’s reaction, when told he had been on a cruise but stuck in an interior room working twelve-hour days, captures the outside view of this kind of focused building. It is weird from the outside. From the inside, it is the only way Spencer knows how to do a project of this scale: go into a cave, eliminate distractions, and do not come out until it is done.
Nobody needs to work twelve-hour days in a cruise ship cabin to learn from this. The lesson for firms is that the tools available right now allow a single person with domain knowledge and AI fluency to accomplish what previously required a team. That changes the math on what is possible, and it changes what firms should be asking for from the people they already have.
2. The CEO Who Does Not Know What to Do
The conversation Spencer had with the CEO is the frame for the rest of the episode, and it is worth staying with for a moment because the CEO’s situation is the norm.
This is a firm Spencer’s team has been working with for more than half a decade, teaching their people real estate financial modeling. The CEO reached out to discuss AI implementation progress, which was not a formal engagement. He opened up because of the relationship. He described a tech team with a plan, an AI harness adopted and deployed, varying degrees of value being generated, and a fundamental uncertainty about what to do next.
Spencer noted that this pattern repeats whether you are a one-person firm or a firm with a dedicated CTO. The pressure is the same. Rates are elevated. Capital is challenged on the debt side. Cap rate compression is not producing NOI growth. Inflation is compressing operating margins. AI looks like the most promising lever available, but most firms cannot articulate what pulling that lever actually requires.
Spencer’s diagnosis: the technology is capable. The problem is the framework. And the framework starts with understanding what he now calls the biggest reskilling and retooling wave probably in human history, bigger in his view than the personal computer, the internet, mobile, the automobile, or electricity. Each of those produced a colossal change. This is probably bigger. And what that means for every firm is that surviving it requires reskilling and retooling, which is a different and harder problem than simply adopting a new tool.
3. Build or Buy Is the Wrong Question
Spencer made a pointed observation about the conference circuit: panel after panel, the question is build or buy. Do we build our own AI systems or do we buy them? Spencer thinks that is the wrong question, and his reasoning is worth unpacking.
The build or buy frame assumes you are making a one-time decision about a discrete tool or capability. You are retooling your entire organization. Every system, every process, every role will look different in ten years. Some of those rebuilds will involve buying external solutions. Some will require building from scratch. But build or buy isn’t a single strategic choice. The real question is how to systematically retool every part of how the firm operates.
He used the example of a CRM. Do you build your own CRM or buy Salesforce? That is a legitimate build or buy question. But it sits inside a much larger question about how the firm manages relationships, how it handles data, how its people interact with digital companions throughout the day. The build or buy decision on the CRM is downstream of the bigger decision about what the firm is actually becoming.
Similarly, do you build your own AI governance policy? You cannot buy one off the shelf. Your organization is different from everyone else’s, your data is organized differently, your people work differently, your risk profile is different. You have to build that. But you might buy the data infrastructure that connects your agents to the information they need. The answer depends on the specific piece, and the framework for deciding which pieces to build and which to buy is the part most firms do not have.
4. The Three Things Firms Getting Ahead Are Actually Doing
Spencer’s three-part framework is the most actionable section of the episode and the one worth spending the most time on.
The first is mapping digital companions into the org chart. The firms that are ahead have methodically identified what work they do, and they have mapped that work across both their human team members and their digital companions. Spencer uses the term digital companion deliberately. He described AI as a new member of the team rather than a tool: one that handles manual, repetitive work that does not require judgment, relationships, or decisions, but that integrates fully into how the team operates and produces output. The firms that have done this have an org chart where the digital companions have defined roles alongside the humans. Everyone on the team, human and digital, has a job description. That clarity is what allows the whole system to function.
The second is building a disciplined approach to structuring unstructured data. Every day, unstructured data lands on the desks of CRE professionals: rent comps in emails, offering memoranda, lease abstracts, T-12s, market reports. Traditionally, that data gets internalized by individuals, who develop expertise over time. When those individuals leave or retire, the expertise leaves with them. The firms getting ahead are building systems to capture that data, structure it, put it in relational databases and vector databases, and connect it to MCP servers so their agents can query it in real time. They are also encoding the knowledge that lives in their senior people into agent skills, so that expertise can be deployed by digital companions long after the individual who originally held it has moved on. Sam’s framing is apt: it is effectively democratizing the collective knowledge of the organization, giving everyone on the team instant access to the accumulated intelligence that previously lived in the heads of a few people.
The third is establishing a practical AI governance policy. Practical is the critical word. Spencer described two examples of impractical policies: a blanket ban on uploading anything to AI because of a legal team’s data privacy concerns, and a mandate that everyone use Copilot while simultaneously restricting access to agent skills, web browsing, and custom connectors. Both of those policies exist in real organizations today. Both produce the same result: analysts doing everything manually that their peers at other firms are having their agents handle automatically, or analysts side-installing Claude on their own laptops and using it anyway in a rogue fashion. A practical governance policy acknowledges the real risks and mitigates against them, while accepting that the organization must move forward. It creates a path for people to build agent skills, connect custom connectors, and implement AI in a way that is supervised and controlled rather than suppressed.
5. Why Copilot Is So Bad and Is Not Getting Better
Spencer was uncharacteristically blunt about Microsoft Copilot in this episode, and the context for that bluntness is worth capturing.
The Copilot critique comes down to what happens when an organization mandates an AI harness that is genuinely not competitive, and then restricts access to the features that would make it more useful. An analyst using a fully featured Claude or ChatGPT environment with agent skills, custom connectors, and web browsing is operating in a fundamentally different world from an analyst using a stripped-down Copilot. The first analyst’s rent rolls are done by an agent. Leases are abstracted automatically. Offering memorandum summaries are generated on arrival. The second analyst is doing all of it manually.
Spencer’s broader observation is that Microsoft has a history of acquiring promising products and making them worse. He cited Skype and Minecraft as examples. His prediction is that Copilot will not meaningfully improve to the point where it closes the gap. The organizations that have mandated it are making a governance decision that has a real operational cost.
The practical implication is what Spencer described as the rogue installation problem. Analysts hear from their peers at happy hours what those peers’ agents are doing. They come back to their desks and run into a policy that prevents them from doing the same things. The ones who care enough install Claude on their own laptops and run it anyway, outside the firm’s visibility and governance. That is a worse outcome than a practical policy would have produced, and it is happening in organizations with overly restrictive AI governance.
6. The Biggest Reskilling Wave in Human History
Spencer has made this claim before across this series, but the CEO conversation sharpens it in a specific way. You can’t opt out of this wave. The analogy he draws is to email and digital calendars. There was a time when using email gave firms an edge. That edge compressed to zero as email became ubiquitous. Now it is table stakes. The same thing is going to happen with AI. The firms that act now get the edge, but the bigger reason is that in some number of years this will be the minimum standard for operating in the industry.
Spencer’s read on the current state: most firms are where this CEO is. They have adopted a harness. They are trying to get usage. They are not yet building digital companions into the org chart, structuring their unstructured data, or operating with a practical governance policy. The gap between where they are and where the leading firms are is real and it is growing. But it is not too late. The organizations that start on these three things now will be meaningfully ahead of the ones that wait another year.
The Bigger Idea
The cruise ship story and the CEO conversation are two versions of the same insight. Spencer rebuilt a major digital asset in nine days because he had the domain knowledge, the AI fluency, and the uninterrupted time to channel both. The CEO cannot figure out what to do next because his firm has the technology but not the framework for channeling it systematically.
The three things Spencer identified, mapping digital companions into the org chart, structuring unstructured data, and building a practical governance policy, are organizational decisions rather than technology implementations. They require leadership clarity about what the firm is becoming, not just what tools it is adopting. The build or buy question is a symptom of not having made those organizational decisions yet. When you know what you are becoming, the build or buy questions answer themselves.
Sam’s framing is the one to carry: if you are a CEO who just got handed an AI mandate and does not know what to do next, this episode is the starting point. The three actions Spencer described are not the only actions worth taking. But they are the ones he consistently sees separating the firms that are ahead from the ones that are still trying to get value from their Claude subscription.
For CRE professionals looking to implement AI in commercial real estate effectively, the framework starts with those three things. The A.CRE Accelerator builds the financial modeling foundation that makes the knowledge layer meaningful. AI.Edge is where you develop the AI fluency to build and manage digital companions. And CRE Agents is what it looks like when someone has already done the work of building the capabilities layer for commercial real estate so you do not have to start from scratch.
Frequently Asked Questions about Episode 19 of Multipliers: Three Things CRE Firms Getting Ahead Are Actually Doing
What are the three things CRE firms getting ahead are actually doing?
Spencer identified three actions based on hundreds of conversations and engagements with commercial real estate firms. First, they have mapped digital companions into their org chart, meaning they have identified what work their AI systems do and how that integrates with what their human team members do. Second, they have built a disciplined approach to structuring unstructured data, capturing the rent comps, T-12s, leases, and market intelligence that flows through the firm daily and making it queryable by agents rather than letting it live only in individual employees’ heads. Third, they have established a practical AI governance policy that acknowledges real data privacy risks and mitigates against them, while still allowing the firm to move forward rather than suppressing AI use entirely.
Why does Spencer say build or buy is the wrong question?
The build or buy frame assumes a one-time decision about a discrete tool. Spencer argues that what firms are actually facing is a complete retooling of their entire organization, every system, every process, every role, over the next decade. Some of those rebuilds will involve buying external solutions. Some will require building from scratch. But making a single strategic call on build versus buy misses the bigger picture: you are figuring out how to systematically retool every part of how your organization operates. The specific build or buy decisions for individual systems flow naturally from having answered that bigger question.
What is a digital companion and how is it different from an AI tool?
Spencer uses the term digital companion deliberately to distinguish AI from the conventional framing of a tool. A tool is something you pick up and put down. A digital companion is a member of the team that has a defined role, shows up in the org chart, and integrates into how the team operates to produce output. The distinction matters because it changes how you think about implementation. You do not implement a digital companion the way you implement software. You define its role, give it the knowledge and tools it needs to do that role, and integrate it into the team’s workflow the same way you would integrate a new human hire.
Why is structuring unstructured data so important for CRE firms right now?
CRE firms ingest enormous amounts of unstructured data every day: rent comps in emails, offering memoranda, lease abstracts, T-12s, market reports. Traditionally this data gets internalized by individuals who develop expertise over time. When those individuals leave, the expertise leaves with them. The firms getting ahead are capturing that data systematically, structuring it into relational and vector databases, and connecting it to MCP servers so their agents can query it in real time. They are also encoding the knowledge of senior people into agent skills. The result, as Sam described it, is effectively democratizing the collective intelligence of the organization, giving everyone instant access to what previously lived only in a few people’s heads.
What is a practical AI governance policy and why does it matter?
A practical AI governance policy acknowledges real data privacy and security risks and mitigates against them, while accepting that the organization must move forward. Spencer described two examples of impractical policies: a blanket ban on uploading anything to AI because of legal team concerns, and a Copilot mandate that prohibits agent skills, web browsing, and custom connectors. Both produce the same result: people doing manually what their peers are having agents handle automatically, or people side-installing Claude on their own laptops and using it anyway outside the firm’s visibility. A practical policy gives people a path to implement AI in a supervised, controlled way rather than suppressing it until the pressure becomes untenable.
Why does Spencer say Copilot is bad and is not going to get better?
Spencer was direct: Copilot as an AI harness is not competitive with Claude or ChatGPT, particularly when firms restrict it further by disabling agent skills, web browsing, and custom connectors. He pointed to Microsoft’s track record with acquisitions like Skype and Minecraft as context for his skepticism that the product will meaningfully improve. The practical consequence is that analysts using a restricted Copilot environment are doing manually what their peers at other firms are having agents handle automatically, which creates both a productivity gap and a talent retention risk as those analysts look for environments where they can actually work the way the technology makes possible.
What does the reskilling and retooling wave mean for individual CRE professionals?
Spencer described this as probably the biggest reskilling and retooling wave in human history, larger in his view than the personal computer, the internet, mobile, the automobile, or electricity. For individual CRE professionals, it means that the way you work is going to change fundamentally over the next decade. The firms that navigate this well will have people who understand how to work alongside digital companions, how to build and manage agent skills, and how to evaluate AI outputs with domain expertise. The professionals who develop those skills now will be better positioned in every environment, whether their current firm moves fast or not.
What did Spencer rebuild in nine days on the cruise ship and what does it show?
Spencer spent nine days in an interior cruise ship cabin, working twelve-hour stretches, to rebuild a major A.CRE digital asset that had previously taken two engineers eighteen months to build. He also leveled up CRE Agents during the same sprint. The same digital asset, rebuilt better than before, in a fraction of the time, by one person with domain knowledge and AI fluency. Spencer described his pattern as going into the cave: blocking out all distractions and going deep on a build project until it is done. The practical lesson for firms is that the tools available right now allow a single motivated person with domain expertise to accomplish what previously required a dedicated engineering team.
What is Spencer willing to do for CRE firms trying to figure out their AI strategy?
Spencer mentioned that he and his colleague Joe work with enterprises on AI adoption, covering the full range from teaching people to work differently with AI, to AI governance policy, to building databases that structure unstructured data and connect to agents via MCP. He also said he is happy to get on a call with longtime A.CRE community members to talk through where they are and how to move their thinking forward. The work they do on the enterprise side is not a fit for every organization, but for firms with meaningful size that have already adopted an AI harness and are trying to figure out what to do next, it is worth reaching out.
What is the main takeaway for a CRE CEO who just got handed an AI mandate?
Stop asking build or buy and start with the three things. Map your digital companions into your org chart. Build a disciplined approach to structuring your unstructured data. Establish a practical AI governance policy. Those three actions are what Spencer consistently sees separating the firms that are ahead from the ones still trying to get value from their AI subscription. They are organizational decisions that require leadership clarity about what the firm is becoming, more than they are technology implementations. The A.CRE Accelerator builds the financial modeling foundation. AI.Edge develops the AI engineering fluency. And CRE Agents is what it looks like when someone has already built the capabilities layer for commercial real estate.
