Large language model (LLM)

The kind of AI model behind ChatGPT, Claude, and Gemini. Trained on huge volumes of text, it predicts the most likely next words given what came before. For a working REALTOR, that means read-time intelligence on the raw text you already wrote.

What it does (the operator translation)

Skip the engineering definition. Here's what matters for a working agent.

You hand the model a wall of text — your last 14 days of messages, a sphere CSV, a listing description draft. It reads everything and predicts what comes next. Sometimes the next thing is a ranked call list. Sometimes it's a follow-up text in your voice. Sometimes it's a comp summary. The model doesn't "know" anything — it pattern-matches against everything it was trained on.

The same architecture sits behind every consumer AI brand you've heard of. ChatGPT runs OpenAI's GPT family. Claude runs Anthropic's models. Gemini runs Google's. Each company tuned the model differently — voice, defaults, what gets refused — but the underlying technology is the same shape.

Andrej Karpathy framed the implication on X in 2025 inside his Software 3.0 thread: English is the new programming language, the model is the runtime. For a 12-deal agent, that's the unfair advantage on owned data — your messy notes have more signal than any structured database somebody else built.

Why a working REALTOR cares (the breakpoint)

For every working agent, an LLM is the most useful general-purpose tool added to the desk in a generation. BiggerPockets practitioner threads converge on the same use cases — listing copy, sphere ranking, follow-up drafts, market summaries. Not magic. Just speed plus voice when paired with a Context Card.

What this is NOT (the category-flip)

An LLM is NOT an AI agent — it doesn't take actions, doesn't fire workflows, doesn't browse on its own unless wired into one. It's a draft-and-think tool. Stapling "agent" branding onto an LLM is what most vendor marketing does — see LLM agent for the boundary.

Where this comes up in The Listing Machine

LLM-as-runtime is the foundational concept in week one of The Listing Machine. Students learn what the model can and can't do before they touch a single prompt — so the rest of the cohort lands clean.