Module 1 — Run your first round on your own work

As of 2026-08-30 this is the first module of 5am School. It exists because most AI education stops at the same place: it explains what a model can do and leaves the reader to work out how that applies to their own job. This module skips the explaining and starts with your own last working day.

What a module is here

A module is three things, always in this order: teaching content short enough to read once, one copy-paste meta prompt that does the integration work rather than describing it, and the resources the prompt produces or points at. No video is required to complete this module — the content below is everything it needs, and a video slot exists in later modules only as an optional addition, never a requirement.

The mechanism, in four steps

This is the same loop every 5am engagement runs, shown here rather than described:

1. A prompt is written about your work, not a generic use case.
2. You run it wherever you actually work — the assistant you already have open.
3. You send back what came out, unedited.
4. The next round is written from what the first one produced.

Nothing here asks you to install anything, connect an account, or send your work anywhere but back to whoever sent you this module. The prompt below runs entirely inside the chat window you paste it into.

Prompt 1 — the four-question intake

Paste this into any AI assistant that can hold a back-and-forth conversation — the tool you already have open is the right one. It asks four questions about one real day, then writes the first version of a prompt for the single task worth automating first.

This is a read-only exercise. You will not open, edit, or create any files — you only read what I paste into this conversation and write your answer back to me in this chat.

I am about to describe one real day of my own work. Your job is to find the exact place a language model could take work off my plate, and to write the first version of a prompt I can start using tomorrow.

Ask me these four questions, one at a time, and wait for my answer to each before asking the next:

1. Walk me through yesterday, hour by hour, at the level of "opened X, read Y, wrote Z." Don't summarize your day — list what you actually touched.
2. Of everything you just listed, which parts are you already doing with an AI tool, even informally? Tell me the tool and roughly how often.
3. Pick the single task from your list that ate the most time and that you understood well enough to check someone else's work on. Describe it in one paragraph: the input you start with, the output you need, and how you know it's right.
4. What do you have to protect — client names, financials, anything you cannot paste into a chat window? I need this so the prompt I write never asks you to paste it.

Once you have all four answers, do the following, in order, and show your work at each step rather than jumping to the end:

- Restate the task from question 3 back to me in one sentence, so I can confirm you understood it correctly before you go further.
- Name the one constraint from question 4 that most limits what this prompt can ask for, and say plainly what the prompt will therefore never request.
- Write a first-draft prompt that does that one task. The prompt should name the exact input it expects, the exact output shape it produces, and one line telling whoever runs it how to tell if the output is wrong.
- Below the prompt, write three sentences: what I should paste in first, what I should expect to see, and the one thing most likely to be wrong the first time I run it.

Stop after that. Do not draft a second prompt, do not suggest a tool stack, and do not estimate how many hours this will save me — you have one data point, which is not enough to estimate anything, and a number produced from one example is a guess wearing a costume.

Resources this module gives you

• The prompt above — copy it whether or not you read it first; the copy button works either way.
• The round log. After you run prompt 1, keep a plain record of what happened. Three columns are enough: the date, the one-sentence task from your answer to question 3, and what the first draft got wrong. That log is what the second round gets written from — nothing here reads it for you, it is yours to keep.
• What comes back to you. The output is a first-draft prompt plus three sentences on how to run it and what to expect. That draft is the resource — this module produces it, it does not hand you a pre-written one, because the one worth having is built from your own day, not from a template.

What this module is not

It does not estimate how much time this saves you. One example is not a measurement, and the prompt above says so on purpose — a number built from a single run is a guess, not a result. It does not ask what you do for a living or who you work for; the four questions are about the work itself. And it does not need a follow-up call to be complete — running the prompt once and reading what it hands back is the whole module.