🛠️ Practice
Time to get hands-on. You’ll guide AI to build real things—from your first app to agents working in parallel—with ready-to-use examples to copy, paste, and run.
0 of 30 topics
The path through Track 2: it starts with a problem and ends with the proof that it worked.
Course roadmap
Detailed content
🚀 Your first guided build
From problem to a working result, without writing a single line of code — just directing the AI.
Pick a concrete pain point from your day (a tedious, repetitive task), not a giant project.
Starting small gives you a quick win and teaches you the process.
Real pain point · small scope · quick win.
Clearly state what you expect to receive at the end — format, style, examples.
AI gets things right much more often when it knows the target.
Clear target · format · example.
Show the AI the reference material: an example of what already exists, its rules, and its data.
Context is what separates a generic result from one tailored to you.
Reference material · rules · data.
The first result is rarely the final one. You point out what to change and ask again.
The magic is in iteration, not the first shot.
Iterate · feedback · refine.
Get it working for real, even if it's simple.
Something running is worth more than a thousand plans.
Working · simple · real.
Make a vague request, want everything at once, and accept the first result without reviewing it.
Knowing the pitfalls saves hours.
Vague request · broad scope · don't review.
🧩 Building with skills
Package your way of doing things into a "recipe" that the AI repeats the same way every time.
When you do the same thing every week and want it to come out the same way every time.
A skill is what turns repeated effort into a single click.
Repetition · pattern · savings.
Every skill says WHEN to use it (trigger) and HOW to do it (the steps).
Understanding the parts helps you write skills that work.
Trigger · steps · clarity.
Write, in plain language, the step-by-step instructions the AI should follow.
You don’t need code—just clarity.
Portuguese · step by step · no code.
Run the skill, see where it makes mistakes, and improve the instructions.
A good skill is refined; it doesn’t come ready-made.
Test · adjust · refine.
Save the skill to use anytime and share it with colleagues.
A skill becomes an asset of your own that pays off over time.
Save · share · asset.
A starter kit: summarize, reply to email, organize, research.
You’ll leave with useful skills in your toolkit.
Starter kit · useful · ready to use.
🔌 Connecting to the World
Connect AI to your apps and let tasks happen automatically.
Map the repetitive tasks that could run without you.
You find out where you save the most time.
Map · repetition · time saved.
Use the “universal outlet” (MCP) to connect AI to email, calendar, and spreadsheet.
That’s what takes AI out of its box and brings it into your world.
MCP · connect · your apps.
What gets the automation started: a time, a new email, a click.
Without a trigger, nothing runs on its own.
Time · event · trigger.
Building a sequence: trigger → AI does something → result goes to the right place.
It's your first real automation.
Sequence · end to end · delivery.
Give the AI only the access it needs and review what it can touch.
Automating responsibly helps you avoid headaches.
Minimum access · review · caution.
Rare, delicate tasks or tasks that require human judgment aren't always worth automating.
Knowing when not to automate is a sign of maturity.
Rare · delicate · judgment.
🤖 Agents that do the work
Delegate entire tasks and have multiple agents work at the same time.
Provide a complete objective and let the agent execute it from start to finish.
That’s the trick: you stop micromanaging.
Goal · delegate · autonomy.
Launch several agents in parallel, each handling one part.
Multiplies the speed of large projects.
Parallel · fan-out · speed.
Break a huge task into parts that each fit one agent.
Smaller pieces deliver better results.
Break · pieces · divide.
Review each agent’s output critically before using it.
Delegating isn’t abdicating — you’re still the conductor.
Review · critical · quality.
Simple tasks = one agent; large, separable projects = several.
Use the right strength for each case.
Simple · separable · your choice.
Coordinate the agents and combine the parts into a single result.
It's conducting in practice—the course topic.
Coordinate · combine · orchestrate.
✅ Verify and prove
Turn what you’ve built into proof: measure it, show it, and build a track record.
Choose, before you start, which number you want to move (time, sales, errors).
Without a target, you can’t tell whether it worked.
KPI · target number · beforehand.
Use the result in real situations to see if it holds up.
"Looks right" can be misleading; testing shows the truth.
Test · real · truth.
Write down what it was like before and what it’s like now—the difference is your result.
It's the evidence that convinces anyone.
Before · after · difference.
A short account: problem → what you did → the number that changed.
A case study is your currency in the AI market.
Problem · action · result.
Publish what you made somewhere people can see it.
Proof tucked away in a drawer opens no doors.
Publish · visible · portfolio.
Do a few more cases to build a track record, not rely on a lucky break.
History builds reputation and prepares you for Track 3.
Repeat · history · reputation.