There is a point where AI stops being a novelty and starts behaving like real operating leverage. Not a “someday this will matter” idea, but a shift in how much meaningful work one person can push through a week. That is the line we talk through in Multipliers – An A.CRE Pod, Season 1, Episode 4: “The AI Crossed the Threshold.”
In commercial real estate, you feel this in the basics. Underwriting capacity jumps from 2 or 3 deals a week to 10 or 20 with the same team. IC memos that used to take days have strong first drafts before lunch. Lender packages, sensitivities, and reporting can keep pace with a live model instead of lagging behind it. At that point, AI is no longer a chatbot on the side of your screen. It is an operator living inside your workflows.
The “Multiplier” idea is simple: some people in every firm already produce 3 to 5 times the value of an average teammate. They see more deals, move faster from thesis to execution, and recycle learning across transactions instead of starting from zero. AI, used properly, lets more people step into that lane.
You can see the before-and-after in a few concrete shifts:
- Intake becomes structured instead of chaotic. OMs, broker blasts, rent rolls, and T‑12s are normalized quickly so you can compare opportunities on equal footing.
- First-pass analysis is no longer the bottleneck. Getting to a credible initial underwrite, with a few sensitivity views, takes hours less time, which lets you say “no” faster and go deeper on the “maybes.”
- Narrative and numbers stay in sync. IC memos, decks, and lender packages pull from the same assumptions as your model instead of being rebuilt by hand every time something changes.
- Institutional memory moves into workflows. The way your team thinks about rollover risk, capex timing, and leverage can be captured and reused instead of living only in a couple of senior people’s heads.
Once that happens, the nature of your work changes. You spend less time copying, formatting, reconciling, and “hunting for the latest file,” and more time on judgment: risk, structure, negotiation, and strategy.
There is a real divide forming here. Firms that only use AI at the edges will see small efficiency gains. Firms that put AI in the middle of their underwriting, asset management, and capital markets processes will quietly lower their cost per decision and shorten cycles across the board. On the personal side, an analyst who can orchestrate an AI-enabled workflow will simply own more deals and more responsibility than someone with the same title doing everything manually.
The path forward is practical, not abstract. Take one workflow, for example “broker email to IC memo,” and map it out end to end. Then ask, step by step:
- Which pieces are repetitive pattern work, and
- Which pieces are true human judgment?
Use AI to handle the pattern work and protect your energy for the judgment calls. Do that across a handful of core workflows and you will feel the threshold moment from the inside: you are not just “more efficient,” you are operating at a different altitude.
Season 1, Episode 4 of Multipliers gets into these dynamics with specific CRE examples and stories from the field. If you want to stay on the right side of the AI divide in this industry, it is worth your time. And if you are already experimenting with AI in your day-to-day, pay attention to where it stops being a helper and starts behaving like a true multiplier. That is the line you want to cross, and then keep pushing outward.
Frequently Asked Questions About AI As Operating Leverage In CRE
Spencer is describing a qualitative shift — not a gradual improvement — in what AI can do and who can use it. The specific inflection he points to is the period around December 2025, when AI coding agents went from unreliable to genuinely functional, even for non-technical users. He frames this with a reference to Andrej Karpathy’s observation that programming had become “unrecognizable” and his own experience watching Michael build something in weeks that would have been impossible six months earlier.
Andrej Karpathy is a co-founder of OpenAI and former head of AI at Tesla — someone Spencer describes as a credible technical realist rather than a hype voice. In late January, Karpathy posted that programming had become “unrecognizable” as a discipline, signaling a structural change in how AI coding tools behave. Spencer cites him because his credibility makes the observation harder to dismiss as marketing.
Michael uses the metaphor to describe a new way of working: a conductor doesn’t play every instrument in the orchestra, but directs world-class musicians toward a shared vision. AI agents are the musicians — highly capable in their individual lanes — and the subject matter expert (the CRE professional, the operator, the analyst) is the conductor. The value isn’t technical ability; it’s knowing what the music should sound like.
Gert, the CTO of CRE Agents, made the observation that the chat interface — opening a tab, typing a prompt, reading a response — is a prototype for the technology, not the technology itself. What Michael has built inverts that model: agentic AI runs continuously in the background, processing data and making recommendations, while chat becomes a feature for interrogating the system’s logic. Most people are still using the prototype.
An agentic AI system operates autonomously rather than waiting for prompts — it processes inputs, makes decisions, and surfaces recommendations without being asked each time. Michael has built one for his outdoor hospitality company that integrates booking data, market signals, review platforms, and accounting software. Each morning, the system has already done the work: pricing recommendations, competitor benchmarks, review responses, and operational flags are waiting for him rather than requiring a manual pull.
An AI skill is a structured set of instructions that encodes how someone thinks, what they know, and how they want AI to behave on their behalf. Spencer calls it IP because it represents the accumulated expertise of a practitioner — deal judgment, analytical frameworks, communication style — transferred into a format that AI can act on. He warned that interacting with public AI models using proprietary methodologies risks giving that knowledge away, rather than encoding it in a system you control.
Marks published a memo in December asking whether AI was a bubble — a skeptical framing. By February 26th, after months of personal engagement, he had shifted meaningfully: he was now more concerned that we were underestimating AI’s potential than overestimating it, and separately concerned that the pace of dislocation might outrun people’s ability to adapt. Spencer uses this arc to illustrate that even careful, measured thinkers are moving decisively on the question.
The threshold Spencer describes isn’t coming — it’s already here. The gap between what AI can do and what most practitioners are doing with it is enormous, and it’s compounding in favor of the people who have started building. The practical path forward is to encode your expertise into AI skills you control, move beyond chat into agentic workflows where possible, and do it alongside other people who are building seriously. AI.Edge and CRE Agents are two places A.CRE has built to help practitioners take those steps without needing a technical background.