Episode 12 of Multipliers: Ask Why Until the Answer Changes

Every industry has a default way of doing things. A standard buy box. A standard funnel. A standard set of tools that everyone uses because everyone uses them. The people who build lasting advantages are the ones who look at that default and ask why. Not once, but repeatedly, until the answer either holds up under scrutiny or reveals that it was never the right answer to begin with.

That is the thread running through this episode. It shows up in Michael Belasco closing a second RV park acquisition using an AI-powered deal screening system. It shows up in Spencer Burton’s elevator analogy for why CRE Agents pivoted its entire product strategy. It shows up in Sam Carlson’s moment watching CRE Agents complete a task from start to finish and realizing, for the first time, that the software was one hundred percent of the way there. And it shows up in the SpaceX conversation, where the first principles obsession that put a reusable rocket on a barge turns out to be the same mental habit that separates the CRE professionals building something durable from the ones doing what everyone else does.

In this episode of the Multipliers podcast, Spencer and Sam cover the CRE acquisition funnel and how AI multiplies it, the difference between a harness and a capabilities layer, why the mental model you use to understand a new technology determines whether you ever actually adopt it, and what SpaceX going public has to do with your buy box.

• You might also enjoy: The prior episode on why most people and companies are ill-prepared for the AI wave: Episode 11 of Multipliers: Most People and Companies Are Ill-Prepared for the AI Wave

• Related: Build the technical foundation this episode references: The A.CRE Accelerator

Episode 12 of Multipliers: Ask Why Until the Answer Changes

Michael Belasco is out again this week, this time closing on a second RV park acquisition in the southeast, which Spencer mentions with the kind of understated pride that comes from watching someone in your orbit do exactly what you knew they were capable of. Spencer and Sam are on their own, and the episode has the loose, generative quality that comes when two people with no fixed agenda start pulling on the same thread from different ends. Spencer is running CRE Agents and teaching at UNC Kenan-Flagler. Sam just finished writing a book on marketing laws and is thinking about funnels in a way that turns out to be directly applicable to everything Spencer wants to talk about. The conversation connects back to Episode 11, where Spencer and Sam worked through the re-skilling and retooling requirement. This one is more conceptual and more personal, a conversation about the mental frameworks that either open up new possibilities or quietly keep you doing the same thing everyone else is doing.

Listen on: Apple Podcasts | Spotify

Why This Episode, Why Now

The episode opens with Michael closing his second RV park and Spencer explaining the CRE acquisition funnel to Sam, which leads quickly into a broader conversation about buy boxes, screening, and how AI multiplies the top of the funnel. That conversation is interesting on its own, but what makes it more interesting is what follows: Sam has just finished writing a book on marketing laws that breaks the funnel into seven stages governed by distinct psychological principles. The overlap between how you acquire real estate deals and how you move a customer through a marketing funnel turns out to be more than superficial.

The funnel conversation sets up the mental model conversation, which is where the episode finds its real depth. Spencer spends time with an advisor who needs everything translated into an analogy before he can engage with it, and the exercise of finding the right analogy for the CRE Agents product pivot produces one of the cleaner explanations of what harnesses and capabilities layers actually are that the podcast has generated. Sam’s Google memory, the moment he first understood what it meant to type a question into a search bar and get an answer, connects to a more recent moment watching Spencer use CRE Agents to complete a task completely, without the usual fifty percent gap, and realizing the mental model had finally updated.

The SpaceX thread is the third piece. The company went public the week of this recording, up fifty to sixty percent. Spencer’s interest is not in the stock. It is in the organizational behavior that made the company possible: a group of people who refused to accept that things had to be done the way they had always been done, and who kept asking why until the answers either held up or revealed that the constraints were not constraints at all.

Episode Highlights

Here are the themes that stood out.

1. The CRE Acquisition Funnel and How AI Multiplies It

Michael’s second park acquisition is the entry point for a conversation about how deal flow actually works in commercial real estate, explained by Spencer with the clarity of someone who has thought about it from multiple angles across a long career.

The funnel is simple: you have a buy box, you screen deals against that buy box, some percentage of what you screen leads to a letter of intent, some smaller percentage of those close. If your hit rate is half of one percent, you need to look at two hundred deals to close one. The implication is that if you can increase the number of deals at the top of the funnel without decreasing the hit rate, you close more deals with the same size team.

AI multiplies the top of the funnel by automating the screening work. The narrower your buy box, the more automatable the screen. Spencer used single tenant retail as his example: if your criteria are a lease length between twelve months and five years, a defined tenant credit profile, and a specific set of markets, that screen can be run programmatically at scale. What used to require someone manually sorting through CoStar listings or broker emails can be delegated to a system that runs continuously and surfaces only the deals that meet the criteria.

Michael’s situation is the live case study. He is a three-person shop with an AI-powered deal sourcing engine behind him, which means he can turn over more stones than a competitor twice his size who is still doing the screen manually. Sam’s observation is the right one: the funnel does not change. The velocity through it does.

Spencer was careful to note this is not new. Before AI, offshoring handled high-volume repetitive work. Robotic process automation handled rule-based digital tasks. Spencer’s own early version of funnel automation was writing broker letters in black Sharpie on a Xerox machine so they looked handwritten and got a higher response rate from farmers. AI is the latest iteration, the most powerful one, but the underlying logic is unchanged.

2. What a Harness Actually Is, and Why It Matters More Than the Model

Spencer’s explanation of the harness concept is one of the cleaner ones he has given, prompted by Sam asking him to break it down for listeners who have not encountered the term.

The large language model is the raw capability: a prediction engine trained on enormous amounts of text that can generate coherent, contextually appropriate responses. On its own, it is not particularly useful to most people. The harness is what wraps around it to make it useful: a user interface, a context management system, a set of conventions for how to interact with the model, and increasingly a set of specialized capabilities that make the model relevant to a specific domain or task.

ChatGPT was the first major harness. It took a capability that existed in research labs and made it accessible to anyone with a browser. Claude, Gemini, and the others followed with their own harness innovations. Spencer’s argument, and the one that has shaped CRE Agents’ product strategy, is that as the underlying models converge in capability, the quality of the harness becomes the differentiating factor. The benchmarks show Opus 4.8, GPT-5, and Gemini 3.1 Pro in a narrow band of performance. What separates them in practice is the harness around them.

The capabilities layer is the piece that most enterprise users end up building themselves, often without realizing that is what they are doing. When you spend an afternoon teaching Claude how to format your quarterly reports, or building a skill that encodes your underwriting methodology, you are building a capabilities layer: domain-specific knowledge and conventions that make the general-purpose harness useful for your specific work. Spencer’s realization at CRE Agents was that this capabilities layer for commercial real estate was actually the product, not the harness they had spent two years building.

3. The Elevator Analogy That Made It Click

The best test of whether you actually understand something is whether you can explain it to someone who has no existing frame of reference for it. Spencer was sitting with a senior industry advisor who had no working knowledge of Claude, no AI.Edge membership, and no particular interest in learning the terminology. Spencer needed to explain why CRE Agents had pivoted from building its own harness to building a capabilities layer that could be delivered through any harness.

The elevator analogy emerged from that constraint. The harness is the building. CRE Agents had been building a building, and it was a fine building. But you cannot get to the seventieth floor without an elevator. So they built an elevator inside the building, and it was a genuinely good elevator. Then they looked around and realized that all these other buildings, Claude, ChatGPT, Gemini, Manus, did not have an elevator designed specifically for commercial real estate. So instead of selling the building, they started offering the elevator. Building agnostic. Installable anywhere.

For the advisor, it clicked immediately. For anyone who has tried to explain the MCP connector model or the skills delivery mechanism through the A.CRE Intelligence Hub, the elevator analogy is worth keeping. It translates the abstract product architecture into something that makes intuitive sense to a real estate professional who has never thought about API integrations or context windows.

The broader lesson Spencer drew is one that runs through several episodes of this series: the ability to turn complex technical concepts into accessible analogies is not just a communication skill. It is a thinking skill. The process of finding the right analogy forces you to understand something well enough to know what it is actually doing, which is a different and deeper kind of understanding than being able to describe it in technical terms.

4. Sam’s Mental Model Shift: The First Time CRE Agents Was 100% of the Way There

Sam’s moment is the emotional center of the episode. He had watched Michael use AI tools for the RV park. He had watched Spencer build and demo things in previous episodes. He was familiar with the technology. And the experience had always been the same: impressive to a point, and then a gap. Fifty percent of the way there. The cool part worked. The rest still required manual finishing.

Then he watched Spencer open Claude, describe a task, engage CRE Agents through the connector, and watch the system work through the pre-built skills for that task from start to finish. No gap. No manual finishing. Complete.

Sam described it as the first time a new mental model fully installed. He drew the parallel to the first time he typed something into Google. There was a moment before the mental model existed, when the concept of a search bar was foreign and the result was surprising. And then there was the moment after, when the mental model was in place and using it became automatic. His CRE Agents moment was the after.

His prediction is simple and probably correct: once most CRE professionals see this once, they will get it. The mental model will install. And then it is just a matter of tinkering: what else can this do for my specific situation? The barrier is the first genuine exposure to the tool working completely on a real task. Everything before that is description. The mental model does not update on description. It updates on experience.

5. SpaceX, First Principles, and the Buy Box Question

SpaceX went public the week of this recording, up fifty to sixty percent on its first day. Spencer is interested in getting in on the stock eventually. But what he actually wants to talk about is the organizational behavior that built the company.

The SpaceX story, as Spencer and Sam frame it, is about a group of people who refused to inherit the constraints of the industry they entered. Every time they were told something had to be done a certain way, they asked why. Not as a rhetorical challenge, but as a genuine inquiry: what is the actual reason for this constraint, and does that reason still apply? In many cases, the answer was that the constraint existed for historical reasons that were no longer relevant, or because nobody had ever bothered to question them.

Sam’s concept of initial conditions is the more precise framing. Every system, every formula, every process was designed with a set of initial conditions in mind: assumptions about what was true, what was possible, and what mattered. When those conditions change, the system that was optimized for them may no longer be optimal. The question is whether anyone is paying attention to the conditions or just running the process.

The application to commercial real estate is direct. Most buy boxes, most screening processes, most underwriting approaches were designed for a world where information was expensive and analysis was manual. Those initial conditions have changed. AI makes information cheap and analysis scalable. The question worth asking, the one SpaceX would ask, is: if we were designing this process today knowing what we know, would we design it this way? For most firms, the honest answer is no.

Spencer’s comment about the buy box conversation he had with Sam that morning is the practical version of this. Sam pushed back on an idea by saying it was not how things were normally done. Spencer’s response: why not? That question, held consistently, is the SpaceX principle applied at the scale of a small CRE operation. It is also, as Spencer noted, something he learned to do early in his career, sending handwritten-looking letters to farmers with a black Sharpie and a Xerox machine because nobody had asked why the standard outreach was not working.

6. The Cost of Intelligence and Why Excelente Exists

The final technical thread of the episode is the cost conversation, sparked by the Cursor acquisition by XAI and the chart Spencer pulled up showing Composer 2.5 performing near frontier model levels at a fraction of the cost.

The chart positions models on two axes: average cost per task and coding competence. The frontier models, Opus 4.8 and GPT-5, sit at the top right: highest competence, highest cost. Composer 2.5, powered by an open-weight model fine-tuned with XAI’s compute, sits at near-frontier competence for under a dollar per task, compared to seven dollars or more for the frontier models. A seven-to-one cost difference at comparable performance is not a marginal improvement. It is a structural shift in the economics of AI-powered workflows.

Sam pays $700 a month for Manus tokens. Spencer uses free models for non-proprietary tasks. The cost of running AI workflows is a real variable that most people underestimate when they think about what it means to go AI-native.

This is the context for Excelente, A.CRE’s free Excel add-in currently working through Microsoft’s approval process. The product was born out of a curriculum need: the new AI-native Accelerator program needed an in-Excel AI tool, and Spencer did not want to force students, many of whom are early-career professionals, to pay for an expensive subscription just to access the curriculum. The team built a model-agnostic harness that accepts any API key and lets the user choose from over 600 models, including the free models that new providers release for testing and training. For non-proprietary work, the cost of AI in Excel drops effectively to zero.

Spencer’s broader prediction is worth noting: harnesses will become increasingly model-agnostic, and users will choose their underlying model based on their own sensitivity to the intelligence-to-cost tradeoff. The platform that locks users into a single model will lose to the platform that lets them optimize. That is the direction CRE Agents is heading, and it is the direction the industry is moving whether or not any particular company chooses to lead it.

The Bigger Idea

The title of this episode is the simplest possible description of the habit that separates the people building something new from the people maintaining something inherited. Ask why until the answer changes. It is what SpaceX did with rocket manufacturing. It is what Michael is doing with RV park acquisitions. It is what CRE Agents did when it realized the elevator was the product, not the building. It is what Sam was doing when he broke the marketing funnel into seven stages governed by psychological laws and found that the conventional funnel story was missing most of what actually mattered.

The mental model conversation is the most important thread in the episode because it is the prerequisite for everything else. You cannot ask productive why questions about a system you do not have a mental model for. You cannot adopt a new tool if your mental model for what that tool can do is built on a prototype version of the technology. Sam’s Google moment and his CRE Agents moment are the same kind of event: the mental model updates, and then the tool becomes usable in a way it was not before.

The funnel insight is worth carrying into any CRE context. The funnel does not change. The hit rate does not change. What changes is the velocity at the top, and AI changes that velocity more than any previous technology because it can screen continuously, without fatigue, against any set of criteria you can define. Michael’s second acquisition is downstream of a system that asks why it should take a three-person team a full week to screen two hundred deals, and finds the answer wanting.

The cost conversation is the one to watch. A seven-to-one price difference at comparable performance does not stay seven-to-one for long. Either the frontier models come down in cost, or the open-weight models improve in capability, or both. The professionals and firms who are building model-agnostic workflows now, who are not locked into any single harness, will adapt to that shift faster than the ones who went all-in on a single platform. Excelente is A.CRE’s answer to that problem for the Accelerator curriculum. The A.CRE Accelerator is where you build the foundation that makes any of the tools worth using. AI.Edge is where you stay current as the tools keep shifting. And CRE Agents is the elevator, installable in any building, that gets you to the seventieth floor.

---

Frequently Asked Questions about Episode 12 of Multipliers: Ask Why Until the Answer Changes