There is a question that changes entire careers: what did you learn today that you did not know yesterday? The founders who got stuck do not answer it well. Neither do the CEOs who have spent ten years repeating the same talks. But the entrepreneurs who keep growing at 60 do answer it, with detail and honesty.
This is the second section of the White Belt in AI Black Belt. After mindset comes learning because a disciplined mindset without fuel shuts down in two years. Learning is what keeps the engine running for decades.
And here there is an interesting turn: in the AI era, the speed and depth with which you learn became the most asymmetric competitive advantage available. Not capital. Not the team. Not the product. The speed of learning. We are going to see it in three planes: business, automation, AI.
1. Learning in business
The book divides the entrepreneur's learning into three categories. Each one requires a different muscle, and each one is trained in a different way.
The good: recognizing the opportunities
It is not only identifying them: it is identifying them before the majority does. The obvious opportunity is already red competition. The opportunity that is invisible to the untrained eye is where the asymmetry lies.
The concrete practice: every week, dedicate 30 minutes to reading outside your industry. If your business is B2B SaaS, read about construction, agriculture, hospitality. The cross-cutting opportunities appear there, not in your industry newsletter. The brain of the mature founder connects patterns between different domains. That is what makes him see what others do not see.
The difficult: facing the challenges
Most founders avoid the difficult challenges. They postpone them, delegate them badly, or resolve them with the first obvious solution to get them out of the way. This is exactly the opposite of learning.
The difficult challenge is the best university available. The uncomfortable conversation with the key client who is leaving. The decision to close a business line you invested three years in. The firing of the first employee, even if it hurts. Each one teaches something that is not learned in books, and that the rest of the entrepreneur's career is going to use.
The practice: when a difficult challenge appears, write yourself five questions that only this challenge can answer, before resolving it. Then, during the resolution, keep those questions as a compass. At the end, write yourself the answers. This is the formal learning of the founder that few people do.
The unforeseen: navigating uncertainty
Plans are never fulfilled the way they were planned. The entrepreneur who learns to navigate uncertainty surpasses those who fear it. This is not resignation to chaos: it is developing the reflex of acting well with partial information.
The practice: in every important decision, write yourself explicitly what you know with certainty and what you are assuming. Then, monitor your assumptions. When one of them is shown to be false, that is information, not failure. The speed with which you react to new information is what separates the agile founders from the stubborn ones.
True mastery does not arrive when you can do it well once. It arrives when it is impossible for you to do it badly.
2. Learning in process automation
When you begin to automate processes, learning changes its nature. You no longer learn only what you do. You learn about how your organization learns.
Unlearning in order to automate
There is a strange phenomenon I see in many clients: the founder knows the process could be done better, but his team defends it because "we have always done it this way". This is the highest barrier to automating, higher than the cost of the tool or the technical complexity.
Unlearning in this context means withdrawing from the team the pride of doing manually what technology can do better. It is not transferring knowledge: it is transferring identity. The assistant who wrote invoices by hand for ten years identifies with that skill. If you automate the process without accompanying the transition, you lose that person, and with her, all the operational context the new system does not capture.
Learning the new bottlenecks
When you automate a process, the bottleneck does not disappear. It moves. It goes from "doing the task" to "supervising the automation," "interpreting the exceptions," "deciding when to intervene manually." Each one of these is a new job your team never did before.
The entrepreneur with learning discipline studies the new bottlenecks before they show themselves. He documents the expected failure modes, trains the team to detect them, and designs the points of manual intervention in advance. The one who does not learn this discovers the new bottlenecks when an important client complains, months after having automated.
Before automating a process completely, run one month in hybrid mode: the machine does 80% but a human reviews 100%. What that human learns about the errors of the system is worth more than the manual of the tool.
Organizational learning accelerates
A good automation does not only free up time: it accelerates the learning of the rest of the organization. When an operational process is documented in code, the new employees learn in days what used to take months. The "operating manual" stops being a PDF that nobody reads and becomes a system that operates and teaches at the same time.
This is a compounding effect that short-term founders never grasp. For them, automation = saving time. For those who understand this point, automation = speed of learning of the organization. That second interpretation yields 10x more in the long term.
3. Learning in the application of AI
Here learning stops being a philosophical principle and becomes the only sustainable competitive advantage available. AI tools change every six months. The one who learns fast wins. The one who clings to last year's mental model loses.
Learning to learn with AI
The central skill is not knowing which tool to use. It is knowing how to learn a new tool in 48 hours. This requires a meta-learning: identifying the common structure between similar tools, mapping the critical differences, and building a fast test to evaluate whether the tool serves your real use case.
This is trained. The first time you learn an AI tool it takes two weeks. The fifth time, two days. The tenth time, two hours. The compounding is brutal, but only if you invest in the deliberate practice of learning new tools, not if you stay comfortable with the one you already master.
Unlearning old cognitive habits
Twenty-eight years writing emails manually formed in me certain cognitive habits: structuring the message mentally before starting, rereading twice before sending, adjusting the tone according to the recipient. When GPT-4 arrived, these habits became partially obsolete. Not because they disappeared, but because the division of labor changed: now the model does the first draft, I provide the context and review. It is a different workflow that requires unlearning the old one.
The managers who resist this are the easiest to identify: they write generic prompts, they read the answers without judgment, and they complain that "AI is not useful for my case". What is not useful is their way of operating with AI. Their mindset keeps them from unlearning the habits that no longer apply.
Relearning the fundamentals in an AI key
The most subtle and the most important. Once you have practice with AI, the fundamentals of your business that you already believed you mastered reappear with new questions:
- What is customer service really, now that an agent can answer 80% of the inquiries?
- What is sales really, when the first contact is made by a model?
- What is leadership really, when the operational decisions are recommended by a system?
These are not problems to "solve." They are questions to relearn the fundamentals in this new context. The founder with learning discipline receives them with curiosity. The one who does not, receives them with defensive fear.
The complete pattern
Learning in business is processing reality with three lenses (the good, the difficult, the unforeseen). Learning in automation is transferring knowledge from people to systems without losing operational context. Learning in AI is developing the speed of learn-unlearn-relearn as a competitive asset.
The entrepreneur who trains the three planes simultaneously has a cognitive compounding his competitors cannot replicate. It is not talent. It is discipline applied to the muscle of learning, sustained over years.
Frequently asked questions
Because mindset without learning runs out. A disciplined mindset without fuel shuts down in two years. Learning is what keeps the engine running for decades. After learning comes the team, because no founder learns alone everything he needs: the team extends his capacity to learn. The three sections form the base without which the later belts do not hold.
Four practices: (1) 30 weekly minutes of reading outside your industry, (2) structured journaling at the close of the day (what I learned, what I would try differently tomorrow), (3) in every difficult challenge, writing the five questions that only this challenge can answer before resolving it, and (4) keeping a circle of peers and mentors with learning discipline. Total: 3 to 5 hours per week. What is learned there surpasses years of formal courses without discipline.
It means withdrawing from the team (including from yourself) the identity associated with manual skills that no longer provide an advantage. An assistant who wrote invoices by hand for ten years identifies with that skill. If you automate without accompanying the transition of identity, you lose the person. Unlearning in automation is not transferring knowledge: it is transferring identity. This is a soft skill, not a technical one. And it is the highest barrier to automating.
Learning a new tool every 30 days with a real use case from your business, not with artificial exercises. Start with ChatGPT/Claude for writing, then Midjourney/DALL-E for images, then Cursor/Bolt for code, then agents (Make, n8n) to automate. Each new tool shortens the learning of the next one because you keep building a meta-model of how these technologies work. In six months, learning a new one takes two hours. In a year, you become someone who can evaluate tools before the majority has heard of them.
The next section of the White Belt
Mindset and learning are individual. But no business is built alone. The third section of the White Belt is the team: how to build a human group that multiplies your capacity for disciplined mindset and continuous learning. Without a team, the individual is a ceiling. With a well built team, there is no ceiling.
Go deeper
Want to bring your team to the next belt?
Book a discovery call or explore the full book.