Lesson 1 · 10–12 min · +25 XP
Get your first useful AI result
Mission 1 of 4
Try the weak version first
Open your AI app and type:
- Sent it and read the answerIt's generic. The AI had to guess who your group is, what the reminder is for, and when.
Mission 2 of 4
Now the strong version
Pick a real message you need this week: a class reminder, a customer reply, a team update. Copy this and fill in the brackets:
Example
- Sent the strong version with my real factsCompare it with the weak version. Same AI, much better result, because you told it what you wanted and gave it the facts.
Mission 3 of 4
Check it before you send it
Read the message once more and compare it with your facts. Did the AI add a time, a price, a promise, or a detail you never gave it?
This happens because AI doesn't look up the truth. It predicts what sounds right based on patterns from huge amounts of text. Usually that's helpful. Sometimes it's confidently wrong. That's why you always review.
- Checked every fact against mineNames, date, time, place, price. Nothing added, nothing changed.
Mission 4 of 4
Improve it
If something is off, say exactly what to fix:
- Asked for one fix and got a message I'd actually send
What you just learned
Your AI added a price you never gave it. Why?
Right. It predicts what sounds right. That's why you check every fact.
Who checks the message before it gets sent?
Yes. You're always the editor.
Lesson 2 · 3–5 min · +10 XP
ChatGPT, Claude, and others: learn the method, not the app
| Company | Its AI (the "model") | Where you use it (the app) |
|---|---|---|
| OpenAI | GPT | ChatGPT |
| Anthropic | Claude | Claude app |
| Gemini | Gemini app |
The company builds the AI. The model is the engine that writes the answers. The app is the screen you type into.
That's how I work. I switch AI models depending on the job, and my instructions and process stay the same.
ChatGPT is…
Yes. OpenAI is the company, GPT is the model, ChatGPT is the app.
A new AI comes out next month. What happens to what you learned here?
Exactly. Learn the method, not the app.
Lesson 3 · 18–25 min · +65 XP
Pakilala Ka sa AI Mo, then build your portfolio
Mission 1 of 5
Fill in your card
Type your answers here. Your card builds itself below, and it stays on this device only.
See an example card
- My card is filled in and true about me
Mission 2 of 5
Put it in your AI app and test it
Most AI apps have a setting for personalization or custom instructions. If yours does, paste your card there. If it doesn't, save the card in your Notes app and paste it at the start of a new chat.
Open a new chat and ask something short, with no explanation:
- The answer already fit meYour business, your audience, your tone. Notice that you didn't have to explain yourself again.
- Saved my card, and the test answer already fit meYour business, your audience, your tone. Notice that you didn't have to explain yourself again.
Mission 3 of 5
Build it with one prompt
Your card is already inside this prompt. Add your name and projects, then send it:
Most AI apps can show the website right in the chat. Look for a Preview or Run option. If yours can't, ask: "How do I save and open this on my computer?"
- I can see my website
Mission 4 of 5
Check it like a visitor
- Clicked every link and button. Each one works.
- Everything about me is trueDid the AI add anything you never told it? Remove it.
- Checked how it looks on a phoneAsk the AI to show how it looks on a phone, or open it on your phone.
Mission 5 of 5
Change one thing
- Changed one thing, and everything else stayed the same
What you just learned
AI can build a website in seconds. Clicking it and reading it is how you prove it's right.
Which one belongs in your card?
Right. Your work and your preferences, never your secrets.
Your portfolio lists an award you never won. What do you do?
Right. Everything on your site has to be true.
Lesson 4 · 10–15 min · +40 XP
Build and change a browser game
Mission 1 of 3
One prompt
Play it a few times. Tick what works:
- Tapping makes it rise
- Passing a gap adds a point
- Hitting an obstacle shows "Game over"
- I can restart
Mission 2 of 3
Describe exactly what to change
Vague requests get vague results. Pick the precise one:
The game is too hard.
Precise. The AI knows exactly what to change.
The game gets boring.
Precise. "More fun" makes the AI guess.
You want a reason to keep playing.
Precise. One clear change you can test.
Now pick one and send it:
- Sent my precise change
Mission 3 of 3
Play again
Did your change happen? Does everything else still work? If something broke, say exactly what you saw:
- My change works, and tapping, scoring, game over and restart still work
What you just learned
AI fixes things better when you describe what you saw and what you want. "It's broken" makes the AI guess. "It doesn't restart when I tap" gives it something exact to fix.
Lesson 5 · 3–5 min · +10 XP
From chatbot to AI agent
What you did in this course
You asked AI for things, gave it context, checked the results, and improved them. You made a message, a portfolio, and a game.
But you still had to check every result, spot the problems, and tell the AI what to fix, one message at a time.
The next level
What that looks like in real work
This is how I use AI every day:
- "Check if my website works on phones and fix what's broken." The agent opens it, finds the problem, fixes it, and shows me proof.
- "Send me a sales summary every morning." It runs on its own schedule, and I just read the report.
- "Read these meeting notes and list who needs to do what." I get a checked list, not guesses.
- "Remember how I like my reports." It keeps my preferences, so I never explain twice.
- One goal, several AI workers: one builds, another reviews, another checks the result.
The rule that never changes
More power needs more control. Even my agents must show proof that the work is done, and they must ask me before sending, publishing, or deleting anything.
The method you learned here (say what you want, give context, review, improve) is the same method that keeps agents safe.
What's next
The next course shows you how to set up your own AI agent: give it memory and reusable instructions, connect it to your tools, and make it prove its work before you trust it.
What must a good AI agent do before sending, publishing, or deleting anything?
Right. More power needs more control.