E1 · Five real requests
15 minOpen the history of your AI chat and copy the summary of five requests from this week into the diagnosis document.
Done when: the table has five rows with requests you really made.
Lesson 1 · 12 min
Three eras: from prompt to role
In just a few years, the way of working with AI has changed three times. First, the request written word by word. Then, the context the AI receives before the request. Now, the role: the AI works like someone on the team and you manage. Knowing which era each of your requests belongs to shows you what's missing.
Start the lesson
In this lesson you will
- 1What each era asked of you
- 2Recognize the era of a request
- 3Why AI became someone on the team
- 4The whole course in five minutes
Goal: sort five of your requests into the three eras and say what each one is missing to reach the role.
01The three eras
Tap or click each era. Notice what changes: who does the work and what you hand over.
Scroll the page: each step below lights up its part of the diagram.
- 1/4
Prompt
The skill was writing the request. Because the model was weak, you made up for it with instructions: "think step by step," "you are an expert," lists of "don't do this." Much of that no longer makes sense with today's models (lesson 2).
- 2/4
Context
The request got shorter because the AI started receiving, before it, what it needs to know: the rules, the files, the history. The skill became building that foundation once, instead of repeating everything in every conversation (lesson 3).
- 3/4
Role
The AI runs on its own for hours or days, like an employee. You don't write what it does word for word: you define the role, the goal and the quality standard, and check in from time to time. This is role engineering (lessons 4 to 7).
- 4/4
From the request to management
With a prompt, you wrote the work. With context, you prepared the foundation. With a role, you manage: you define, check and correct.
One era doesn't erase the one before it: the role uses a clear request and good context. What changes is where your work is.
02What each era asks of you
Tap or click each era and compare the same request.
1. Prompt
You describe step by step how to do it, because the model couldn't figure it out on its own.
The result depended more on how you worded the request than on your knowledge of the subject.
Typical request
You are an award-winning copywriter. Think step by step. Write a post about our course. Don't use emojis. Don't be wordy. Use a friendly tone.
2. Context
Who the audience is, what the company tone is, what has already been published: this is kept in a place the AI reads before answering.
The request stays short because the rest is already there.
Typical request
Write the launch post for the new course, following the tone and audience in the rules file.
3. Role
You hand over a whole job, not a single task: what success is, how to measure it, where the context is, what good work looks like.
The agent works, checks itself and shows you the result. You review and correct, as you would with someone new on the team.
Typical request
You take care of promoting the courses. Every Monday, publish one post per new course, following the example post's format. Done when: each post has a link that opens and has been approved by me.
03Template: the diagnosis of your requests
Copy this into a document (Word, Google Docs) or a spreadsheet and fill it in with five requests you made to the AI this week.
Diagnosis (document or spreadsheet)
# Diagnosis: which era are my requests in? | # | request (summary) | era | what was missing to reach the role | |---|---|---|---| | 1 | <request> | prompt / context / role | <measurable goal? saved context? example?> | | 2 | | | | | 3 | | | | | 4 | | | | | 5 | | | | Questions for each row: - Did I write HOW to do it or WHAT success looks like? - Did the AI already know the context, or did I repeat everything? - Could it check on its own whether it was finished?
Simulation: illustrative example
- "Summarize this PDF in 5 points" · prompt · it didn't say who the summary is for or what to do with it.
- "Write the email in the tone from the rules file" · context · it was missing the done-when criterion (who approves, by when).
- "Every Friday by 5 p.m., put together the report just like the example; done when the totals match the spreadsheet and I approve" · role · it has a measurable goal, an example, a deadline and who approves.
- Thinking a long request is a role request: length doesn't change the era. What changes it is having a measurable goal, saved context and an example.
- Skipping the "what was missing" column: it is your plan for the next lessons.
04Practice: diagnose your requests
Use real requests from your week.
E2 · The era and what was missing
15 minFor each request, mark the era and write what was missing using the three questions from the template.
Done when: all five rows have the era marked and the "what was missing" column filled in.
E3 · The role candidate
5 minPick the request you repeat every week. This is the one that becomes the role sheet in lesson 4.
Done when: one of the requests is marked as "role candidate", with its frequency noted (e.g., every Monday).
05The whole course in video
Five minutes: from the shift to AI as an employee through to the cost.
Forget the prompt: in 2027, you manage the AI (video in Portuguese) · 5 min 11 sVideo page with script →
06Test what you learned
The explanation appears after your answer.
Question 1 of 3
What changes the most from the prompt era to the role era?skill
Correct
In the role you manage: goal, context, example and checking. The request may even get shorter. Switching tools doesn't change the era.Not that one
In the role you manage: goal, context, example and checking. The request may even get shorter. Switching tools doesn't change the era.Review: The three erasQuestion 2 of 3
"Write the email in the tone from the rules file" is in which era?
Correct
The request uses a saved base (the rules file): that is context. To be a role it would need the measurable goal and the routine, such as who approves and when.Not that one
The request uses a saved base (the rules file): that is context. To be a role it would need the measurable goal and the routine, such as who approves and when.Review: What each era asks of youQuestion 3 of 3
With the role era, what happens to context?
Correct
One era doesn't erase the one before it. The role builds on context (lesson 3); what changes is where your work is: defining, checking and correcting.Not that one
One era doesn't erase the one before it. The role builds on context (lesson 3); what changes is where your work is: defining, checking and correcting.Review: The three erasQuiz complete
You got {a} of {n} right. If you get them all right, the lesson is marked as completed.Choose an answer. The explanation appears after you try.
07The lesson in three sentences
Three erasPrompt (writing the request), context (preparing the base) and role (managing the AI like someone on the team).
One supports the otherThe role uses a clear request and good context. The difference is that you hand over a job, not a loose task.
The diagnosis is the planWhat was missing in each of your requests is what the next lessons teach.