Trail map
Detailed content
💥 The 10× question
Problem before the tool: the question that reveals the bottleneck, the numbers that prove it, and the choice of a single process.
“If the business grew 10 times tomorrow, which process would break first?” The question exaggerates the workload so the weak point becomes visible.
An agent executes the plan quickly, whether it's right or wrong. Choosing the problem matters more than choosing the tool.
Process, bottleneck, the 10× question, problem before the tool.
A filter: if no one does that for a week, will the business lose a customer, money, or a deadline? If so, break it down. If not, it's just annoying.
The list of complaints mixes the two things together. Automating an annoyance wastes effort and opens a door without a payoff.
Breaks, annoyance, this week's test, queue.
During a typical week, track how much time each candidate takes, how much money is lost, and how often things go wrong.
Your guess usually points to what’s visible. The number points to what costs you—in Marcos’s case, 10 hours per week.
Time, money, error, estimate labeled.
Common selection mistakes: automating what’s fun or visible (posts, videos, little bots) instead of what quietly costs you.
Every agent is another door to control. And bringing in more customers for a process that leaks only makes the leak worse.
Cost × visibility matrix, good and bad signs, data leak.
A ready-to-use template to paste into an AI chat: it interviews you, one question at a time, until it finds the process that's breaking and the number that proves it.
It’s easy to dodge the question on your own. The chat keeps you focused—and you talk about the process, never the customers.
Interview, one question at a time, don’t make up numbers.
Four steps to decide: make a list, put numbers to it, cut what affects money, and write "I'll start with X because it costs Y."
One agent at a time is the only way to know whether it worked and stayed under control.
Just one, a waitlist, a low-cost error, a decision statement.
🏥 A request isn't a problem
Renata’s clinic case: behind the request for “more clients” was a schedule that was losing R$ 1.080 per month. From pricing the pain to the first small project.
The request is the ready-made solution you’re given (“more clients”). The desire is what’s behind it (“more revenue”). The pain is the loss you can count.
Following the request to the letter can be costly and still not solve the problem. Asking “what for?” leads to the number.
Request, desire, pain point, "what for?".
Renata wanted more patients through Instagram. The schedule showed something else: patients who booked and didn’t show up.
More people joining a broken schedule only increases no-shows. Fix the leak before opening the faucet.
Absences, paid hours without revenue, a leaky bucket.
Absences × average value × weeks. At the clinic: 9 absences × R$ 120 = R$ 1,080 per month, almost R$ 13,000 per year.
Without a price, the pain is just an opinion. With a price, it becomes a decision—and later, the metric for track 5.
How many times, how many fail, how much can be recovered, the cost of the pain.
An agent that sends a reminder two days ahead, asks for confirmation, and records the response. Rescheduling and fitting people in go through Renata.
The first project needs to fail cheaply. In the worst case, one extra reminder goes out.
Small project, timeline, approval, low-cost error.
The same pattern in other businesses: the request points outward, while the pain is inside. Marcos asked for speed; the pain was overdue documents turning into fines.
Recognizing the pattern in other people’s businesses helps you see it in your own.
Sounds like a request × sounds like a pain point, first project by profession.
“Today we lose ___ because of ___ because ___; the first agent will ___.” A template to review with an AI chat.
It's the first field in the agent's profile and the starting point for the Track 3 controls.
Pain statement, number, cause you can change, small action.