🎯 Intent is the destination
Intent is the result you want to achieve with the solution — the destination that guides every decision afterward. Which channel to use, which agent to create, which tool to connect: everything is decided by looking at the intent. It is neither the task nor the tool; it is the destination. And without a clear destination, any path seems good — because you have nothing to compare it against.
🧭 Intent as the destination
Think of a trip. Without a destination, you drive well, burn fuel, cover miles—and don't get anywhere in particular. With a destination, every turn has a criterion: does it bring you closer or take you farther away? Intent does the same for your AI solution. It turns "doing things with AI" into "achieving a result with AI."
✓ With intent
- ✓Each choice has a clear criterion.
- ✓You can say “this doesn’t serve the goal.”
- ✓The result can be recognized.
- ✓The effort converges in one place.
✗ No intent
- ✗Any automation seems like a good idea.
- ✗Nothing seems insufficient—or sufficient.
- ✗You switch tools without a clear reason.
- ✗A lot of activity, little direction.
Key concepts
❓ What we want to solve
The first question of intent is to name, precisely, the problem the solution will tackle. The most common mistake is confusing the desire to “use AI” with a problem. “I want to use AI in customer service” isn’t a problem—it’s a vague desire. “The customer waits 4 hours for a simple answer and gives up” is a problem. The difference? The second one hurts, is concrete, and can be checked to see whether it was solved.
Vague × specific — write yours
If the sentence doesn't fit in one clear line, it's still too vague.
💡 Practical tip
To test whether a problem is well defined, ask: "who is affected by this, and what changes when we solve it?". If you can't answer in one sentence, the problem is still vague—and a solution to a vague problem solves nothing.
Key concepts
👤 Who we want to help
Every solution serves someone specific: the end customer, the support representative, the manager, the business owner. Each audience changes everything—the response tone, the right channel, what counts as success. A solution for the customer needs to be simple and fast; the same goal, aimed at the manager, needs reporting and visibility. Trying to solve things “for everyone” effectively means solving them for no one.
End customer
Wants a quick, simple answer.
Support agent
Wants to serve more people without getting overwhelmed.
Manager
Wants a clearer view and better decisions.
Owner
Wants results and growth.
🎯 Who feels it × who decides
Often, the person who feels the pain and the person who decides on the solution are different. The support agent feels the overload; the manager decides to invest. Knowing who you’re solving the problem for—and who needs to see the benefits—determines how the solution is designed and presented.
Key concepts
🏆 Which result matters
Before building, define what the "before" and "after" — the concrete result that proves the solution worked. This is the antidote to the most common trap in AI projects: “it looks nice, but nothing changed.” A defined result ties the solution to a real gain, not an impression of modernity.
Describe the “before”
What things are like today, without the solution: how much time, how many errors, how much waiting.
Describe the “after”
What will it look like when it works? That’s the result you’re working toward.
The difference is the gain
The leap from "before" to "after" is the value the solution delivers.
✓ Concrete result
- ✓"Answer in 5 min, not 4 hours."
- ✓"80% of questions resolved without a human."
- ✓"2h a day returned to the team."
✗ Vague impression
- ✗"It looks more modern."
- ✗"People thought it was cool."
- ✗"Now we use AI."
Key concepts
📏 How We’ll Measure
Measurement isn’t the last step—it starts along with the intent. While you’re designing, choose the number that will tell you whether it worked: time saved, customer requests resolved, errors reduced, recovered sales. If you don’t decide on the metric at the start, you end up measuring whatever is easy at the end—and almost always fool yourself with numbers that mean nothing.
The intent metric—fill it in
If you can’t fill in "valor_hoje," you don’t understand the problem well enough yet.
💡 Practical tip
Measure intent, not effort. "I ran 200 prompts" is effort. "Response time dropped from 4 hours to 8 minutes" is intent. The number that matters is the one that shows the problem being solved—not how much you worked.
Key concepts
🧭 Clarity as the first skill
The most valuable skill for an intent architect isn’t technical—it’s think clearly. Define the problem, audience, outcome, and measure before touching any tool. Tools change every week; clarity is what makes any tool deliver. It’s the only asset that doesn’t become obsolete—and it’s what separates those who deliver from those stuck chasing the next new thing.
Problem, audience, outcome, and metric—in writing, not just in your head.
Only then do you decide on channels, agents, and tools—guided by intent.
The tool is the part that changes the most and matters the least at first.
🧭 The asset that never ages
Those who learn a tool master a tool. Those who learn to think clearly master every tool that comes along. A well-defined intention is what makes the rest of the course—architecture, prototype, measurement—work. Without it, every subsequent module becomes a disconnected attempt.
Key concepts
Self-check (optional): why does intent come before the tool?
🎯 Module summary
Next module:
3.2 — Map the Problem: understand the company and see where AI can multiply results.