🎯 Skate to where the puck is going
In hockey, a good player doesn’t chase the puck. They watch who has it and anticipate where it will be, so they can get there first and, with luck, score. Products work the same way.
The practical question: if the industry is heading toward a point where your idea could be swallowed up, will your customer notice? And when?
✓ Those who skate to where the puck is going
- ✓Read the signals: new pain points, falling costs, changing habits.
- ✓Ask what will become easy in 3-6 months and what will stay difficult.
- ✓Build the part that won’t become a commodity.
- ✓Arrive when the customer starts looking.
✗ Those who chase the puck
- ✗Copy what’s already succeeding today.
- ✗Build on a limitation the next model will solve.
- ✗Arrive when the market is already crowded.
- ✗Compete on price alone.
💡 A simple habit
Every week, write down three things that have become easier with AI. In a month, you’ll have a list of what’s about to become a commodity and what hasn’t yet.
🙈 Arbitraging customer naivety
Many people won’t discover that they can vibe-code a particular solution for another six or twelve months. Until then, there’s an arbitrage on market naivety: you sell something they could technically build themselves, but they don’t know that.
The plan for when the window closes
Scale
Grow enough for your brand, data, and distribution to protect you when the naivety wears off.
Sell
Negotiate a sale of the business (or the data) while it’s still worth a lot.
Pivot
Use your customers and what you’ve learned to tackle the next problem that hasn’t become a commodity yet.
⚠️ Arbitrage isn’t deception
Customers pay for convenience, trust, and time saved. That’s legitimate. Hiding information or exaggerating how difficult something is to keep them locked in isn’t.
🛢️ Slow-moving industries mean longer windows
Think of industries like oil and gas, precious metals, mining, traditional agribusiness. Even if a model ten times better came out tomorrow, many of them would just be discovering that you can connect an AI assistant to email.
These companies make a lot of money by doing one thing well in the physical world. They’re focused on metallurgical processes, logistics, and workplace safety, not on which model launched this week. If even people who work in tech can’t keep up with everything, imagine people who work in other fields.
✓ Why it’s worth targeting slow-moving industries
- ✓They have plenty of money and costly processes.
- ✓They won’t build it themselves.
- ✓They value relationships and support.
- ✓There’s less startup competition.
✗ What they’ll require
- ✗Longer sales cycles.
- ✗Industry language, not tech jargon.
- ✗Security, compliance, and contracts.
- ✗Patience for implementation.
💡 Translate, don’t impress
In these industries, nobody wants to hear “agent with a frontier model.” They want to hear “the report that used to take two days now comes out in twenty minutes, in the same format as always.”
🐭 Cat and mouse versus saturation
Think about interface component libraries for developers. Most of the content is free: you copy a prompt or a code snippet, and you’re done. There are hundreds of alternatives, many open source. Why would anyone pay? It’s hard to make money from a problem that’s been solved well enough and has spread everywhere.
| Question | An answer that supports recurring revenue |
|---|---|
| Does the problem keep coming back? | Yes, every week or whenever the environment changes |
| Does the solution need continuous updates? | Yes, the world changes and the solution keeps up |
| Are there equivalent free alternatives? | No, or they’re much worse |
| Does the customer lose something if they cancel? | Yes: history, protection, data, routine |
That’s why the most useful question is: what recurring pain is worth solving that most people won’t solve on their own? Not the sexy problem, but the boring one that keeps coming back.
🔄 Be part of the daily workflow and generate passive progress
It’s very hard to build a product that stays in the background and that users want to have there. Two examples from the world of software development show the way.
Error monitor (e.g., Sentry)
It watches the application and alerts you when an error shows up for users. Developers don’t have to remember to open anything: the alert comes to them.
Code reviewer (e.g., CodeRabbit)
It analyzed a huge number of code changes and learned patterns in errors. After the agent writes the code, it warns you: “this will break in production; have it fixed.”
Notice the second example: it lives in symbiosis with the tools automating development itself. A lot of people don’t even look at the code the agent writes anymore; they “close their eyes and pray.” The product makes that prayer safer. It doesn’t compete with automation; it builds on it.
💳 Example: bank statement becomes an investment thesis
Imagine an app where you connect your bank account through a provider with security certifications. Every day, it reviews your spending and suggests investments that align with what you consume, with an explanation.
If you pay for almost everything with a credit card and notice that everyone does the same, it makes sense to consider payment companies. The app doesn’t invest for you: it gives you a list and explains why, using data you already had but weren’t putting to use. What we buy reflects what we value, and that often resembles what similar people value.
⚠️ Financial recommendations require care
The example is illustrative. A real product in this area must comply with local regulations (in Brazil, CVM and the Central Bank) and make clear that it isn’t individual financial advice.
🧸 A good feeling and becoming a verb
Remember when running antivirus software was a ritual? The progress bar moved, “no threats found” appeared, and you felt relieved. That warm feeling of security is part of the product. A boring topic like security can deliver a very good feeling.
⚠️ The dark side: manufactured demand
There are old reports of vendors helping create the problem so they could sell the solution. That’s not strategy; it’s fraud. The legitimate lesson is different: identify a real fear and provide real peace of mind.
The last signal is the rarest: can the product become a verb? “Google it,” “get an Uber,” “send it on Slack.” Nine out of ten products never get there, and that’s fine. But it’s worth asking whether the name and main action are simple enough to become part of everyday speech.
Copy and run
Assess how recurring your idea is using the signals in this module and get concrete suggestions for increasing it.
Assess how recurring my product is.
Product: <what it does, in 2 sentences>
Audience: <who uses it, industry, technical level>
Expected current usage frequency: <daily / weekly / monthly / occasional>
For each signal below, answer “yes,” “partially,” or “no,” and explain in 1 sentence:
1. Cat-and-mouse problem (renews itself)
2. Naivety arbitrage (how many months until the customer knows how to do it on their own?)
3. Slow-moving industry (long window)
4. Lives in the daily workflow / runs in the background
5. Creates passive progress using data the user already has
6. Provides a good feeling (relief, security)
7. Potential to become a verb
Then suggest 3 concrete changes that would increase usage frequency without adding friction.💡 Combine the signals
The strongest products combine several signals: a security monitor for AI-built apps is a cat-and-mouse game, runs in the background, brings relief, and keeps accumulating data. Each signal reinforces the others.
🧪 Quick module quiz
Three questions. Click an option to see the answer.
1. What is “naivety arbitrage”?
2. Why do sectors like oil and gas offer long windows?
3. What makes an automated code reviewer a good example of a recurring product?