PTENES
INEMA.CLUB PRO Terminal and Harness

AI for decision-makers · 6 short lessons

The chat responds.
O terminal runs.

For executives and independent professionals who already use AI in the browser and still copy and paste everything. Understand the four layers of an AI system, why the terminal with an agent is the one that gets the work done, and how the harness makes it work your way.

get started now →

15 to 21 min per lesson · nothing to install · read on your phone or computer

1

browser chatgreat for one text at a time; can't see your files

2

AI in the editorsees the open file; made for people who code

3

terminal with an agentreads the folder, runs it, checks it, and fixes it: it has hands

4

harnessrules, permissions, and recipes for it to act your way

In 2 minutes

The layers, on video

From chat to harness, and where everything lives in the folders: why the terminal is the layer that executes. Watch before you begin. Vertical version for Shorts/Reels.

What you’ll know how to do

Results, not theory

Place any AI tool on the map in one of the four layers, with its advantage and the point where it gets stuck.

Read a real agent session in the terminal and point out where it read, ran, requested permission, and checked.

Write the fixed instructions your agent’s: three work rules and one limit, in Portuguese.

Decide task by task what stays in chat, what goes to the terminal, and what goes into the harness.

Draw the project folders: one per project, the .claude drawer, the open AGENTS.md, and the knowledge base (wiki and OKF).

"Chat responds. The terminal executes. The harness makes sure it executes your way."

Who it’s for

Made for decision-makers, not programmers

it's for you to

  • You run a department, an office, a clinic, or your own business, and want to know what AI actually does beyond answering questions.
  • You already use ChatGPT or Claude in the browser and feel that big tasks are still manual.
  • You want a clear argument for deciding where your team should work with AI.

it's not for you if

  • You already code and want to learn commands and configuration in depth (look for the Claude Code technical track on INEMA).
  • You want a course with a long video lesson: here, the video is 2 minutes, and the rest is reading and practice.

Learning path

From chat to harness

1

The chat responds, and stops there

Where the browser chatbot helps and exactly where it gets stuck, using one of your own cases.

17 min
2

The four layers, side by side

Chat, editor, terminal, and harness on a napkin, and why the terminal has hands.

17 min
3

The terminal in practice

A real agent session, line by line: read, planned, asked permission, executed, checked.

17 min
4

The harness, piece by piece

The seven parts, and the first version of your agent’s standing instructions, written.

21 min
5

Which layer for which task

The four-line decision rule applied to five tasks from your week.

15 min
6

The folder is the agent’s office

One folder per project, the .claude drawer, AGENTS.md open, the knowledge wiki, and the Google Cloud OKF format.

22 min

see the full path →