Get up to speed on any topic before a call

No materials, no expert on speed dial. Just a chat window and the right prompt.

What you'll have when you're done

A reliable method for getting oriented on any topic you do not have background in, fast enough to be useful before a meeting. You will know the prompt that frames the conversation correctly, how to deepen it in layers so the orientation actually holds, and what to verify before you carry anything into a high-stakes room. Plain chat, no terminal, no setup.

Forty-five minutes out

Here is a situation I have been in more times than I want to admit. Board call at 10. Topic is something I have never run operationally: a legal structure question, a technical architecture decision, an unfamiliar financial instrument, a market I have never competed in. The prep deck arrived last night. I read it and understood roughly half of it.

The old move was to call someone. If I had the right contact and they picked up, I came away better. If I didn't, I'd walk in and manage on instinct, which works until it stops working.

The new move is a fifteen-minute chat session that gets you to a usable frame before the call starts. Not expertise. Orientation. There is a difference, and knowing which one you need is most of the battle.

This is the learning version of meeting prep. If you already have the materials and just need to process them fast, that is a different workflow: Prep for a meeting by pasting in the agenda. What follows is for when you are starting from scratch on a topic, not preparing materials you already have.

What you need first

• A Claude or ChatGPT account. Plain chat window. No plugins, no terminal.
• Fifteen minutes. Enough to run two layers deep and still have time to jot what you learned.
• The name of the topic. That is all. You do not need materials. The whole point is that you don't have them.

That is it.

Step by step

The framing prompt

Most people open a chat and ask something like "tell me about [topic]." That prompt produces an encyclopedia entry. It is organized for a student. You are not a student; you are a decision-maker with forty-five minutes.

This is the prompt that works:

"Explain [topic] to me like I am a CEO who has to make a decision about it. Assume I am smart but know nothing about this specific area. Start with the 30-second version, then the things I would be embarrassed not to know, then the questions I should be asking."

Why it works: it sets your register as a decision-maker, not a student. It states your prior knowledge honestly (smart, but unfamiliar), which calibrates what the model skips and what it explains. And it structures the response in three usable layers you can consume in order, or cut off when you have enough.

"Tell me about X" gets you a textbook. This prompt gets you a briefing.

The progressive-depth ladder

Do not skip straight to tactics. Layer up.

Layer 1 (broad): the 30-second version. Read what comes back and ask yourself: does the orientation feel right? If something in the first paragraph is backwards based on what you have heard in passing, say so. Correct the frame before you go deeper. A Layer 3 recommendation built on a wrong Layer 1 is worse than no recommendation at all.

Layer 2 (deeper): the live debate. Once the frame holds, push on it:

• "Go deeper on [the biggest risk you mentioned]."
• "What are the two camps on this? What is the live debate?"
• "What would an expert assume I already know that I don't?"

The third question is worth running every time. Experts have a shorthand insiders take for granted. That shorthand is exactly what you will miss in the room without knowing why you missed it.

Layer 3 (action): your meeting questions. Only run this layer once the frame and the depth hold.

• "What are the five questions I should ask in this meeting?"
• "What would a skeptic push back on, and how would I respond?"
• "If I had to make a decision today, what is the single most important thing to understand first?"

The last question forces a ranking. You want to know what is doing the most work before you sit down.

For more on how to follow up within a conversation to go deeper, that article covers the mechanics of building on an initial response without losing thread.

Use cases where this pays

A board or investor call on a topic you are weak on. Run Layer 1 and Layer 2 the night before. Walk in with the frame and the two camps. That is enough to hold a peer conversation.

A vendor or technical evaluation. Layer 3's "questions I should ask" prompt turns a vendor demo into a structured evaluation instead of a guided tour.

Entering a new market or category. Layer 2's "live debate" question gets you the map fast, including who thinks the move is a bad idea and why.

Interviewing for a function new to you. Hiring a CFO or a VP Engineering when you have never run that function yourself. Layer 3 gives you questions a domain expert would recognize as real.

How you'll know it's working

You stop getting caught flat-footed. The first meeting where you use a term correctly that you had never heard two hours earlier, that is the signal.

You ask the question that reframes the room. The "two camps" question at Layer 2 tends to surface the frame the other side is operating from. When you name it in the meeting, it lands differently than a question that comes from prepared materials.

You separate settled from contested. You will start noticing when someone in a meeting is stating a contested position as if it were settled fact. That awareness comes from running Layer 2. It is worth more than the orientation itself.

When it breaks

This is the section to take seriously, because it is the one that will cost you if you skip it.

The normal catch is gone. When you use AI on something you know well, you notice when it is wrong. You have ground truth. Here you have none. The model can be wrong, confidently, in a way you cannot detect from the output alone. That is a different and higher-stakes version of the usual verification problem.

The full verification habit is worth reading once. For this specific use case, four practices matter most.

1. Ask it to flag contested vs. settled. At the end of Layer 2, ask: "What here is well-established vs. actively debated?" The model will often surface the difference clearly. Anything it marks as contested should be handled that way in the meeting. Do not carry a contested position as if it were fact.

2. Ask where to verify. "If I wanted to check the most important claims here against a primary source, where would I look?" A useful model gives you a real starting point: a regulator's website, an industry body, a canonical paper, a named report. An evasive answer, or a string of confident suggestions that don't quite exist, is a warning sign. For patterns to watch for when a model is generating plausible-sounding information that doesn't hold up, that article covers the tells.

3. Cross-check anything load-bearing before you repeat it as fact. If a number, a regulatory claim, or a technical assertion is doing real work in what you plan to say, open a real source and confirm it before you say it. Not a second AI. A primary document, an industry site, or a person who has actually operated in this space. This step takes five minutes and has saved me from real embarrassment.

4. Flag recency risk. Ask: "Flag anything here that may have changed in the last six months." Fast-moving areas, including regulation, AI infrastructure, and market structure, shift faster than a model's training data. The model may have a confident view of a rule that was updated last quarter. How AI works and why it has a training cutoff explains why the model does not know what it does not know. Spot-check anything it flags as potentially stale.

Do not present AI-learned facts as your own settled knowledge without running at least one of these checks. The orientation is useful. The unverified orientation presented in a high-stakes room as if it were hard-won expertise is where things go wrong.

Level up

This workflow is for learning mode: no materials, plain chat, fifteen minutes. Once you have run it a dozen times, two natural upgrades show up.

The first is combining it with materials you do have. When you have a deck and a topic you are weak on, run the framing prompt first to build the frame, then switch to pasting the materials and running the prep questions from the meeting-prep workflow linked above. The orientation makes the materials legible; the materials make the prep specific.

The second is using the model to pressure-test the decision itself once you have enough context to have a position. The Layer 3 "what would a skeptic push back on?" question is the on-ramp.

Start with the framing prompt. Run the layers in order. Verify before you carry anything into the room. The whole session is fifteen minutes, and you will walk in a different kind of ready.