A PREVIEW OF THE BOOK

Building High-Performance Teams with Agile Methodologies and Artificial Intelligence

Roberto Canales Mora

You are reading a free sample from Building High-Performance Teams with Agile Methodologies and Artificial Intelligence, by Roberto Canales Mora. It includes the introduction and the first chapter through section 1.5, offering a substantial sample from the start of the book.

The complete book continues with the methods, examples and relationships that help us build self-organising high-performance teams.

English cover of Building High-Performance Teams with Agile Methodologies and Artificial Intelligence
English sample · Edition 0.12.56

Free English sample · v0.12.56: the introduction and chapter 1 through section 1.5, with eleven figures and cited references.

0. To prosper or to live better

In this chapter

We’ll start with a question I invite you to ask yourself: if we spend so much time working, do we want to prosper, live better, or both?

I’ll connect experiences, books, and models so you can weigh what is useful to you and apply it to your own situation.

You’ll also see why I begin classes by asking for volunteers to search, share, and learn together from the very first minute.

For years I’ve taught at business schools, working with professionals who have already spent a long time making decisions. It frustrates me—and perhaps other teachers too—that I never have enough time to cover everything I’d like. The programmes are intensive, there are many instructors, and each gets only a few sessions. Besides, we aren’t robots: every group moves at its own pace and opens up different conversations.

In one class I share something interesting, in another something else comes up, and in a third an example appears that I would have liked to explain to the earlier groups.

That’s why I make myself write from time to time. This book brings together what I usually talk about and what I wish I could develop more fully in the executive education programme. Here I can finish an anecdote, connect it to another one, and add nuance to something that I state categorically in class to wake us up, but which needs more room to be understood. I hope you find this reading useful whether you’ve been in my classroom or found your way here on your own.

Even so, if you get the chance, I recommend coming to class. Each version of the book is a snapshot of what I’ve learnt and managed to organise up to that point, while the teaching keeps changing.

Technology advances enormously, and teachers never stop learning either. Sometimes we even get tired of hearing ourselves repeat the same things, so we look for a different way to explain them, try a new tool, or include something that has just happened to us in a company. Therefore, the classes I teach are usually richer and more current than what I manage to leave written here.

Besides, it’s rare that I teach the exact same course twice. You only have to look at the students’ faces to realise that an explanation hasn’t landed, or that an example hits close to home and deserves more time. Attendance, the group’s pace, the number of questions, and the problems students bring up all change what happens in the classroom.

A teacher can pause, ask you to tell your story, and adapt the conversation; that interaction produces lessons I can’t predict when I prepare a session or finalise a version of the book. That’s why I like you to use these pages to prepare for the programme, review it, and keep thinking afterwards, and to use the in-person class to compare notes with me and your classmates about what is really happening to you.

I’ll also keep updating the book. If you buy it on Leanpub, you’ll be able to download the new versions I publish from your library, at no additional cost. Check the version and date on the cover, both in the book and in the sample. I’ll name the tools I use because they help explain how I work; when they change, I’ll revisit those references. What you read today should help you act now and reconsider how you work when new possibilities appear.

For years I’ve given away thousands of books and tutorials. This time I’ve decided to set a price close to 100 euros. I’m not asking you to buy on impulse: I want your decision to be deliberate, for yourself and in recognition of the work behind these pages. After that, you’ll have to choose a practice and try it; the price alone doesn’t change anything.

I don’t teach these courses so people can leave knowing how to recite a methodology. Methodologies are a means; what interests me is for professionals to prosper in their companies or live better. If you spend a substantial part of your life working and are also making the effort to attend a business school, it makes sense to ask what you want to get out of all this. Being exhausted, getting home late, and still feeling just as stuck five years from now doesn’t sound like the best life plan either.

That’s why I usually start by offering two fairly radical positions: either you aim for your professional life to go very well and become wealthy, or you aim to live well. If you manage both, excellent. I want us to stop using effort to justify any outcome and start thinking about how we organise our working lives, because working hard does not, by itself, prove that we’re working in the right direction.

When I say wealthy, I mean having sufficient assets that give you the freedom to choose, not simply having an expensive car and a collection of monthly payments that force you to keep running, scared every month that you might lose your job.

To give you a concrete reference, Capgemini classifies people with at least one million dollars in investable assets, excluding their primary residence and certain personal property, as high-net-worth individuals (HNWIs) [18]. Earning a high salary and spending all of it every month is something else. A townhouse in a nice neighbourhood, two cars, children at a good school, and thirty-year debts can look great.

But if you live in fear of losing your job because you couldn’t keep paying, how much freedom do you have left? That’s what Robert Kiyosaki popularises in Rich Dad Poor Dad with the image of the rat race [33]: running faster and faster to keep up with what you’ve already committed to. Consider how many months or years you could live without working, whether you could take a sabbatical, or risk starting a business without your life collapsing economically.

Living well can be understood as spending more time with your family, being able to take a few days off without the department falling apart, or working on projects that interest you. And here’s the sentence I want you to remember. In both cases, you have to do exactly the same thing: build self-organising high-performance teams. If everything needs you standing right there, you face an obvious limit, both to growth and to enjoying what you’ve built.

Throughout the programme, I’ll connect ideas from books, business experiences, counters we turn over on a table, and conversations with my children. It might seem like we’re changing topics, but often we’re seeing the same behaviour from different perspectives. If you always do your children’s homework for them, always assign tasks to your subordinates, and always answer the questions, you’re training other people to need you. Don’t be surprised later when they won’t leave you alone.

0.1. Grasping ideas and building useful models

Before we begin, I want to tell you how I’d like you to read this book. I’ll speak with strong conviction and sometimes provoke you a bit; you don’t have to agree with me on everything. I want you to grasp these ideas: listen to what I’m saying, connect it to what you already know, test it, and keep what you can use.

If you’ve been working for twenty years, your judgement on many things may be as good as mine. I’m not asking you to abandon it; I’m asking for a little humility as you look for what you can transfer from someone else’s experience to your own situation, even when their company or way of speaking is nothing like yours. If an idea makes you uncomfortable, try to understand what I’m trying to express before deciding whether you agree.

I often recall a thought from Bertrand Russell: “In the modern world the stupid are cocksure while the intelligent are full of doubt” [1]. It applies to me too: having run a company or explained something many times doesn’t turn my opinions into universal truths. I’m sharing my experience and my criteria so you can build your own. With that agreement, we can move forward without stopping at every example to list all its exceptions.

Figure 0.1. Grasping ideas with discernment: a reflection by Bertrand Russell
Figure 0.1. Grasping ideas with discernment: a reflection by Bertrand Russell

Since I come from computing, I explain it in bits. If I had only one bit in my head, when positioning myself on a topic, I’d have to choose between zero and one. If I ask you whether you’re right-wing or left-wing, you could only pick one of the two extremes. However, if we had eight bits, we could choose from 256 values, from zero to 255.

If the left were zero and the right were 255, you could answer 17, 120, or 144; it would be rare for you to always respond with zero or 255. So let’s try to make use of those degrees. I don’t need to agree with every idea from a political party, every page in a book, or every practice in a methodology.

I heard Javier Recuenco express the idea that you shouldn’t always buy all your fish from the same person. Even a fishmonger who usually treats you well might one day be interested in getting rid of a fish that isn’t right for you. We can trust and still use our judgement. The same goes for a political party: if I accept and defend everything its leader says, whatever it may be, I’m not showing much independent judgement; I’m turning into an easily led disciple.

Figure 0.2. Listen, connect, test, and adapt: there are many nuances between zero and one
Figure 0.2. Listen, connect, test, and adapt: there are many nuances between zero and one

Recuenco also made me think of the well-known phrase by George Box: “All models are wrong, but some are useful” [17]. That’s what I want to do in class: build heuristic models that help us succeed, even if they aren’t valid in all situations. A heuristic guides a decision when I don’t have all the information; afterwards, I see what happens and adjust my judgement.

It’s worth remembering that things aren’t black or white. I can tell you, “stop assigning tasks,” and then spend several pages explaining how I help people find them, what context they need, and when I step in. If you take only the slogan away, you can get a disastrous result; if you strip it of all its force, nothing will change. The point is to understand it and try it, thoughtfully, in your own situation.

I’ve been teaching agile methodologies for years, but it would make no sense to keep teaching the same class I taught two years ago as if artificial intelligence didn’t exist. Some practices now let us prepare in minutes what used to take days to put on the table; others still require conversation, judgement, and agreement among people. That’s why AI appears from the beginning and runs through the whole journey.

I want us to use agile methodologies and AI to build a culture of self-organising high-performance teams, able to deliver value without having the person in charge solve everything for them.

This handbook preserves that journey so you can return to it after the programme. First I’ll talk about developing our own capabilities and how we learn; then about the value and relationships that make a team work. We’ll look at the models I use to think about work and a game that lets us test them. We’ll follow Lean principles, get into Kanban, and end by tackling projects we don’t yet know how to define fully. Scrum and XP will have their place once we understand what problem we want to solve.

I don’t want you to finish the book with a collection of new names and an equally full calendar. I want you to find things you can do, things you can stop doing, and people with whom you can build something lasting. We’ll return to that idea often, because prospering with people you enjoy working with has a lot more staying power than winning an argument on Tuesday morning.

0.2. Learning together from the very first minute

At the end of each chapter, I’ve gathered the ideas I want you to take away as laws: a first sentence to help you remember them and a second to put them to work. I invite you to read them slowly and ask yourself which one you could try with your team. They are numbered consecutively throughout the book, so we can refer back to one without repeating it in full every time.

There’s an activity I propose at the beginning and that we’ll repeat throughout the course. I ask, “Who’ll do it?” and I want someone to say “I will” before they even know whether they’ll have to find a book, check a reference, or ask AI for an explanation. The task will be small, and someone else will take the next turn. The person who does it reviews what they found and shares it with a little context in the class WhatsApp group; a small individual effort benefits everyone.

We can also ask someone in each session to collect the important phrases and concepts. I don’t need the whole class to focus more on copying than listening. If we rotate that responsibility and organise our notes into a map at the end, we all get a resource that would have taken one person much longer to prepare. I encourage you to do this in other teachers’ classes and in your departments too. We’re already practising initiative, sharing the work, and managing knowledge before we study a methodology.

Figure 0.3. Small shared efforts help knowledge reach everyone
Figure 0.3. Small shared efforts help knowledge reach everyone

Later, when I ask you to look up what muda means or find techniques for preparing a meeting without wasting time, we’ll be using the same activity. And if you’re reading the book on your own, you can look it up then, save a note with its source, and think about whom you’d like to share it with. You don’t have to wait until you finish the book to start building a team.

I’ll use a mountain to help us find our way. The summit is a team that delivers value, learns, and organises itself without needing the person in charge at every step. The camps are numbered in reverse, like the steps still ahead of us: Camp 6, personal judgement and capability; Camp 5, value and relationships; Camp 4, work that flows; Camp 3, autonomy and coordination; Camp 2, business and projects; Camp 1, checking value and sustaining improvement.

Trust, collaboration, and continuous improvement appear at the base of the illustration because we’ll practise them throughout the climb: sharing a question, keeping an agreement, or correcting a way of working together are all ways to start building them. We keep learning and improving at the summit; self-organisation also means making sound decisions without constantly consulting a paternalistic leader who decides everything for the team. Teams need to develop judgement, capability and responsibility, within the limits set by risk.

Figure 0.4. Six camps that bring us closer to a team that delivers value, learns, and organises itself
Figure 0.4. Six camps that bring us closer to a team that delivers value, learns, and organises itself

0.3. What we take away to build the team · Laws 1–3

Law 1. Methodologies are a means; the end is to prosper or live better. Before choosing a practice, be specific about how it will help you deliver more value and build a team that can work without depending on you at every step.

Law 2. Grasping an idea means testing and adapting it, not swallowing it whole. When you hear a categorical statement, look for the problem it points to, compare it with your experience, and try the part you can use.

Law 3. A small shared search gets the team learning from the very first minute. At your next session, rotate who looks up a reference, add the source, and share a sentence of context with the group.

Figure 0.5. Why learn together
Figure 0.5. Why learn together

The first capability I want to build is a shared sense of purpose and the confidence to contribute from the start. If everyone waits for instructions even for a small search, it will be difficult for them to organise complex work with others later. That’s why the course begins by practising initiative and judgement before naming the methodologies.

Now that we’ve agreed on why we’re here and how we’ll learn, I want to start with our own capabilities. It would be difficult to ask someone to change how they work if we aren’t willing to review our own way of working.

1. Empowering ourselves first to empower others

In this chapter

We’ll start with our own capabilities: adopting tools and learning to work with AI, while applying judgement to data and results.

I’ll tell you how I use dictation, automation, and a logbook to make better use of my time and what I learn.

Then we’ll look at how to get more out of a book and share knowledge so it reaches the team too.

1.1. Updating our software and working with AI

The first thing we need to do is develop ourselves so we can develop our teams and the people we work with. I often say that we have hardware into which we need to keep installing software: we’re still ourselves, but we need new knowledge, new judgement, and tools that let us do things we didn’t know how to do before. If all your time goes into repeating what you already know, where will that new capability come from?

Figure 1.1a. Camp 6: capacity, judgement, and shared knowledge
Figure 1.1a. Camp 6: capacity, judgement, and shared knowledge

Artificial intelligence needs to be a priority tool at work right now. I don’t understand how a professional who wants to prosper can fail to make a conscious effort to learn to use it, much less how a manager can fail to help their team do so. “I’ll get to it when I have time” doesn’t work for me: we’re talking precisely about gaining the capacity to work better. When I prepare a presentation, an explanation, or an analysis, my first question is how AI can help and what I need to check to make good use of its work.

Before choosing a tool, let’s remember why we want it. We need to deliver value to our customers and our organisation. An architect can design a house on their own; designing a city requires many architects and many capabilities working together. When we think of Apple or SpaceX, we remember one person who embodies leadership, but there are whole teams behind them.

One person has limits. They might start something extraordinary, but they need other people to share the practices and know how to carry on without always waiting for the founder or the boss. So if you want to go far, it’s almost certain you’ll have to learn to lead more people.

You’ll have to learn things you don’t know yet; they’ll handle areas you can no longer attend to personally. In the age of AI, that doesn’t mean teams need to grow without limit: they can also achieve outstanding performance with these tools.

Suppose I need to prepare a presentation for my team. I can explain to AI what we need to decide, who will be at the meeting, what data I have, and what format would work for me; I’ll always use information I can input into that tool.

I ask for a first draft and review it: if a number has no source, I check it; if it suggests something our team can’t do, I explain that and we adjust.

Then I use the presentation to talk and make decisions with people. That journey—context, first draft, review, and use—is what the figure shows. I’m still responsible for the result.

Figure 1.1. From context to use: a first draft needs review and a person to take responsibility
Figure 1.1. From context to use: a first draft needs review and a person to take responsibility

1.2. Dictating, walking, and revising this very book

Writing this book serves as an example for me. I use a Plaud device to record my explanations and get a transcript [34].

Then I work on that material with Codex, the AI assistant I use to prepare and review this book, listen to the result in audio while walking, check if it makes sense, and dictate corrections in my own words.

Sometimes I pause for several minutes and then continue. When I get back, I add those notes to the project and review what has changed. This way, I’ve spent many fewer hours in front of the screen, and I don’t have to choose between taking a walk and moving the book forward.

I still need to read the book through in the usual way at the end and review every sentence, but I’ve done most of the work more flexibly and away from the screen.

It costs money, of course, but my time costs money too. Naval Ravikant, an entrepreneur and investor known for his reflections on wealth and happiness, suggests setting a fairly high aspirational hourly rate and delegating tasks that cost less than that amount [2]. That rate is an aspiration: it forces me to think about where I want to focus my capacity and how much I’m willing to pay to get my time back.

Imagine arguing with your partner over weekend chores, when someone could come to your house several times a week for an amount that might not be decisive for your finances. Do those arguments make sense? Wouldn’t you use that time better by exercising or training? For me, right now, time has a lot of value.

In my case, the device has removed a very specific friction (burning out my eyes on the screen or printing hundreds of pages multiple times) and lets me write while I walk; you’ll have to find what task is stealing that time from you.

Another discovery that has been a lifesaver for me is Wispr Flow [34]. I use it to dictate into text fields on my computer: I turn on dictation, explain what I want to write, and the tool prepares the text, even when I correct myself as I speak. When you see how clumsy I am at typing and how many mistakes slip through, you’ll understand why I find it useful.

I’m less reluctant now to write an article or an email that explains things properly. In fact, I can dictate the idea to an AI so it completes the content or form with the context I provide. Then I check names, figures, and meaning, because a word can be spelled correctly and still be the wrong word.

Figure 1.2. From a spoken idea to a version I can review and share
Figure 1.2. From a spoken idea to a version I can review and share

In class, I provoke the students a little: “You should all be constantly using ChatGPT, or an equivalent tool, as your first option when preparing work.” It’s useful to have it at hand and know its basic features, within your organisation’s conditions. It’s quite possible that part of your team is already using one, with or without your knowledge. Not knowing that your employees use it doesn’t make the usage disappear; it just means you know less about how work is changing.

1.3. Understanding where data goes and getting started with automation

One of the first things we look at is data controls. In a personal ChatGPT account, I ask students to find the “Improve the model for everyone” setting and understand what it means to have it turned on or off [4]. If I’m going to enter company information, I need to know how it may be used. A personal account and a tool subscribed to and configured by the company can have different confidentiality terms.

Business plans offer specific commitments; OpenAI, for example, says it does not use data from its business products to train its models by default [4]. That doesn’t replace corporate policies on what information we can share, which should be clear to everyone. If you need more control over AI use at the enterprise level, another option is to run models inside your organisation.

In that case, it’s important to see the problem as a whole and check where the data goes: an application can run locally and still have connectors that send information to other services. Ask someone who knows, and give your team a guideline it can follow. But unilaterally banning AI use without a parallel training plan seems like business suicide today.

I’m surprised by organisations that prohibit the use of external services and are doing nothing to experiment with internal alternatives. Imagine a programmer who still copies and pastes code as they did five years ago, or who uses a personal phone because the company gives them no reasonable option. They’re wasting the resources technology offers us today.

They produce much less per unit of time than someone who has the right tools at their disposal. It’s like comparing digging a trench with a pickaxe and shovel to doing it with a mini-excavator. Perhaps a local model, tested on a specific task, would help a lot. If you’re running an organisation and neither understand these possibilities nor have anyone studying them, I think you’re leaving an important responsibility unattended.

To get the conversation started, I’d distinguish three uses. First, assistance: I ask it to help me review a confidentiality agreement, point out what worries me, and check the result before sending it to whoever has to decide. Next, automation: when updating this book, I can prepare routines that transform files from each format via a sequence and review the result before uploading it to Leanpub.

Then there’s an agent I assign to find a device with certain features and prepare the purchase; I confirm the payment. In all three cases, there’s work I still do as a person, but it’s not the same work as before.

To move towards useful automation, I choose a frequent task, measure how long it takes now, define what I consider a correct result, and try an alternative. I could generate invoices for a course based on registrations and review exceptions, prepare a draft response to similar emails, or turn my dictation into a first spreadsheet or presentation. I count review time, the autonomy I grant, and errors too.

If an agent can take action, I agree on what it can do on its own and when it needs to ask me to decide. That way I don’t celebrate the tool producing things faster when the work has simply shifted to whoever has to correct it.

Figure 1.3. Assistance, automation, and agents: three ways to expand our capabilities with AI
Figure 1.3. Assistance, automation, and agents: three ways to expand our capabilities with AI

We can also be creative to gain capacity: one possibility is to work with universities. I’d approach professors in industrial organisation or computer science and build a relationship so we can collaborate on undergraduate or master’s theses, or applied research. They need real problems, and we need to explore capabilities we haven’t mastered yet.

If I bring together someone with recent knowledge of AI or methodologies and someone who knows our operations well, both can learn: one understands the business and the other discovers what they might do differently. They need guidance and an agreed scope of work.

To build that relationship, I’ll have to free up time for someone or delegate it to someone capable of looking after it; it won’t work if I email a professor and disappear.

In fact, if I learn to delegate tasks and empower my subordinates, I can devote part of my schedule to finding opportunities that will benefit all parties. Moreover, those people coming from universities could be our best pool of workers.

1.4. A logbook that brings back what I want to learn

When I browse X or other social networks, I find articles on AI, psychology, sociology, or management that interest me. They used to end up in a collection of links shared on my Discord channel that I never reopened. So I asked my assistant for a skill—a reusable set of instructions—to turn a source into a clear, educational explanation with references and an infographic.

In other words, I created a skill in Codex that activates when I say “create a logbook entry from this link.” That’s how my logbooks at robertocanales.com began [34]. They aren’t an inventory of what I already know; they capture what I want to learn, prepared in a format that’s easy for me to consume and review. First there’s a curiosity or a need of mine; then I create the material that helps me understand it.

Later, seeing how useful they were, I asked for an audio version and a player that would let me listen to entries one after another like a podcast. Now I can give the assistant a link, say what aspect interests me, review what it prepares, and revisit it while I’m walking, travelling, or going to the gym. I make use of a moment when I might otherwise never return to that content.

I listen to get closer to the idea; if I want to learn how to apply it, I go back to the text or try it out. If I don’t like how the explanation turned out, I ask Codex to rebuild it with a different approach or more infographics, and in a few minutes, I have another text and another audio file. Plus, the material is saved: it doesn’t get lost like a note in a notebook might. Anyone following my logbooks can share my interests or enrich me with ideas and links I hadn’t thought of.

When my daughter started her entrepreneurship studies, I wanted her to be able to use the same setup to learn on her own. I bought her a domain, arranged the hosting, and passed her the instructions so her assistant could build her own logbooks at patriciacanales.com [34]. I like that an improvement I’ve already worked on doesn’t have to be reinvented by her.

Imagine doing the same thing within a department. A question someone answers today can become a shared note with its source, an example, and a review date. This collection of external knowledge, connected and retrievable, is often called a second brain, an idea Tiago Forte develops in Building a Second Brain [64].

Accumulating information isn’t enough on its own: we need to know what’s worth keeping, how to find it, and who maintains what changes. For this book, we’ve also prepared a concept map to explore relationships, not just folders.

Andrej Karpathy describes a proposal like this in LLM Wiki: keep the sources and, with help from AI, maintain linked notes that grow richer as you add information and answer questions [41]. I don’t want the work of connecting ideas to disappear when a conversation ends. We have to review those connections and distinguish what the source said from what we’re interpreting.

Iván Carrillo is putting a lot of work into this idea of a Second Brain, both explaining it and building a solution to put it into practice. Here’s a video about his proposal, and if you want to know more, you can visit hardment.com [62]. I’d like you to see different ways to organise knowledge and choose the one that helps you retrieve it, connect it, and share it with your team.

We can bring this way of saving valuable knowledge collaboratively and processing it with a second brain (an AI with advanced instructions) to our department or company. Better yet, we must do so. Imagine an everyday situation: we chose a supplier because they could deliver in two days, even though another was cheaper. If I only save the supplier’s name, the next person in charge may replace them thinking they’re improving purchasing.

If I keep the need, the alternatives, the reason we chose them, and when we should review it, someone else can understand the decision and change it with good judgement. A note about that supplier could be connected to delivery time, the cost of a shutdown, and the quality we need. That looks like knowledge that accumulates to me, and that both people and machines will be able to process in the future.

Figure 1.4. Keep sources and reasons so you can revisit a decision
Figure 1.4. Keep sources and reasons so you can revisit a decision

When we start working this way, we’ll want to see how that information relates within the company. That’s where knowledge graphs appear: how does what my team learns connect to knowledge elsewhere in the company? Each connection on the map should let us open the note and understand why it links two things and what decision it helps us make. If it only looks attractive, it still needs work.

I also want to find contradictions: perhaps a recommendation that worked for us a year ago no longer fits our current volume. I’d save the earlier decision and the reason for changing it, so the assistant doesn’t turn two criteria from different moments into a confusing answer.

A student asked me whether these tools would reduce the size of teams. I can speak especially about software development: someone defining a product can now prepare much more concrete proposals and prototypes, while integrating, maintaining, and checking what has been built will still require specialists.

If I produce faster, I also have to review faster. The same goes for a campaign or a document: speeding up creation shifts work toward definition and validation and, when we connect systems and handle information, cybersecurity. This is security in the sense of protecting systems, access, and data, not safety in the sense of making sure a machine doesn’t injure anyone. We’ll have to watch where the new workload appears.

I can give you an example with this book. Writing it has cost me little because I can record with Plaud while walking or dictate to the computer with Wispr Flow. I can review “the big stuff” while listening to the result and dictate corrections with Plaud. But no one takes away the six hours of narration I have to listen to for each regeneration (even if I can play it back at double speed), nor the full reading before a new publication. If I outsourced the publishing, the reviewer would have more work, not less.

One study can help us think better about that question. In The Cybernetic Teammate, Dell’Acqua and his coauthors studied professionals at Procter & Gamble working on innovation challenges. In the study published in 2025, the average quality of proposals from people working with AI was comparable to that of proposals from pairs working without AI, and the gap between technical and commercial perspectives narrowed [42].

I find that very interesting, although we have to look at it with caution because they were observing one working day, not the months it then takes to see a product through. Still, it gives me clues to think that in certain tasks we could approach doubling productivity. In other jobs, the multiplier could be even greater; we’ll have to measure it in our context.

In another experiment, published in 2023 and carried out with BCG consultants, using AI improved work on some tasks and reduced accuracy on another that fell outside the model’s capabilities. The authors called this a jagged technological frontier [43]. Anyone who has used AI can verify that it solves some tasks excellently and others not so well; it even seems to “have days.”

At the same time, its capabilities improve rapidly and continue to surprise me. It’s no longer enough to dismiss it by saying it “hallucinates” or “isn’t reliable,” as we did recently; nor can we stop checking its proposals, even with another AI. I’d ask my team to test tasks of our own: summarising a request, preparing alternatives, and checking a conclusion may not work equally well.

I’d save one example where it works and another where it fails, along with the check that helps tell them apart. That way we stop arguing about whether “AI is good or bad” and start finding out what we can rely on it for in our work.

We should also save the date of the study and the model used. Technology changes quickly, and results from a few weeks ago can become obsolete. A test I did six months ago might come out better today; another might still fail. Plus, we have to periodically ask ourselves what other tasks we could automate. What seems like an achievement today, because we’ve just started, may soon seem basic and encourage us to turn the screw on automation again.

Still, I’m going to warn you about one thing. At first, discovering what you can do with artificial intelligence can be addictive. But it also has its less pleasant side: it’s tiring to spend the day in front of the screen waiting to see what it proposes. I myself spent weeks putting many hours into my master’s project with AI.

Being able to define, design, build, and evaluate like a one-person army doesn’t mean I want to do it indefinitely. Once the novelty wears off, I get tired too, and I want other people to carry the work forward. A growing product needs customer service, marketing, support, and the ability to handle several changes at once. That’s why this book’s goal remains to build a culture of self-organising high-performance teams, even as their size and tools change.

1.5. Reading so knowledge can find me again

There are times when the day-to-day grind doesn’t leave you time to step back and look at where your life is heading. Are you sure your actions are leading you where you want to go? Do you even know where you want to go? Do you at least know where you don’t want to go? Imagine that a copy of you writes from five years in the future and tells you what they wish you had started doing today.

Perhaps they’ll ask you to take better care of yourself, learn another language, or stop waiting for your company to decide when it’s your turn to train in artificial intelligence. What’s interesting about the exercise is that you probably wouldn’t be very surprised by what they said; often we already know what we’re putting off.

There’s no need to turn that conversation into a huge project. I’ll keep saying it: what matters is direction, not speed. It doesn’t make much sense to run without knowing where we’re going. If we always frame things in extremes—going to the gym every day or doing nothing, earning a degree or never learning, changing the whole company or resigning ourselves—we’re making it hard on ourselves.

We can set aside some time to walk, read, or explore a tool that might help us at work, and see what changes we can sustain. But we’ll have to make room in our schedules; otherwise, the letter from our future selves will remain empty words.

It’s worth asking ourselves what the point is of spending hours watching shows, playing games, or scrolling through videos on our phones, only to say later that we have no time to learn. I’m not going to organise your leisure time, but I am going to ask you to do the maths. If you can find time for everything else but never for your future, you’ve already decided what place your professional progress occupies.

I do some of that updating through reading. I still like physical books, and years ago I stopped worrying about having several on the go: one at home, another in my suitcase, another in the village. I can pick them up whenever I get the chance, because my goal with a management or skills book is to make good use of it, not to finish it quickly just to add it to an annual reading count (a vain number). For that, it helps me more to return to some ideas than to finish it very quickly.

My system is quite simple. I do a quick first read: I care more about identifying interesting paragraphs than assimilating all the content at once. When I find something that interests me, I fold down a page corner without marking the exact paragraph. If I stop reading for a while, I leave a different mark so I know where I am.

When I finish, whether it’s been a week, a month, or a year, I open the folded pages and try to find again what caught my attention; that little effort makes me read and think about that idea again—an idea that has already passed through my mind twice.

Then I write a summary in my own words and try to connect the ideas I’ve chosen; that effort requires more elaborate processing. If I publish it on my website or a professional network, I review it again to make sure it’s clear and fix any potential mistakes I missed.

If someone comments on an idea months later, I have to go back to the summary I wrote, recover the context, interpret what that person has said, and strive to respond coherently. That forces me to retrieve ideas that had likely weakened in my memory. That way, the content I found useful has passed through my mind several times, and I’ve also had to explain it.

Now imagine keeping that habit up for years: don’t you think I’m likely to consolidate ideas and find connections between things I used to see as separate? That’s the kind of accumulation I’m after, not a bookshelf proving how fast I read. What you need to ask yourself is whether you get that much value out of books. You may recognise the idea of reinforcement learning here.

Figure 1.5. Reading, reviewing, writing, and sharing feed future learning
Figure 1.5. Reading, reviewing, writing, and sharing feed future learning

I’ve also learnt that the same book can mean very different things depending on when you read it. It happened to me with a book that brought together strategy models: the first time I read it, I thought it explained almost nothing. After taking a business strategy course, I returned to it and found the diagrams very useful—diagrams I hadn’t known how to interpret before.

The book was the same, but I had a different context for reading it, and in my head, it went from seeming like rubbish to seeming like a marvel. The book hadn’t changed; I had.

I do something similar with audiobooks and podcasts. They let me get close to a topic while walking or travelling, but listening doesn’t prove that I’ve internalised it or that I know how to apply it. To tell you the truth, I used to get much less out of audiobooks; since I got Plaud, I can take notes in almost any situation with barely any effort.

If something matters to me, I go back to the text, check it against other sources, or try it out. I can also ask AI to build an explanation tailored to a specific question; the logbook we just looked at helps me keep that learning from getting lost among hundreds of links.

When I think about how we internalise knowledge, both individually and as teams, I return to this phrase: having ten years of experience is not the same as repeating the same year ten times. There’s a question I like to ask people who have spent a long time doing the same work: have those years given you different experiences, or have you repeated an experience you already knew many times?

In class, I illustrate it with a bus driver on a circular route. On the first day, they learn the vehicle and the route; during the first year, they get to know the variations for holidays and usual bad weather. After ten years, have they acquired ten years of different experiences, or have they repeated the same year ten times? The tasks on that route can be quite limited.

In an intellectual profession, however, can we ever stop learning? I find it difficult, but learning requires conscious interest. I don’t only want to know how long you’ve been in the job; I want to know what you can do now that you couldn’t do before.

A friend’s son told me one day that he wanted to talk to me. I responded with the phrase “when the student is ready, the teacher appears”: he would have to consciously seek me out and persist, because sometimes we don’t value what we get without effort. Perhaps he isn’t aware that he has me at his disposal almost whenever he wants, while other people would pay hundreds or thousands of euros for that same time.

If he sought me out, he had to be willing to put in his part: interest, work, and some faith in the process. If not, neither of us would benefit from the conversation. He hasn’t come yet; I suppose that if the day comes, he’ll be ready.

When I share what I’m reading with the team, I also want us to build shared criteria and interests. I make sure they see the books and even “shove them in their faces.” As one employee said: “When you learn, we all learn, whether we want to or not.” I’ve never considered that a negative thing.

If you know which books I’m reading, how I interpret a problem, and what ideas change my mind, it will be easier for you to anticipate my judgement. And if you share yours, you help me too. That conversation builds a more aligned way of thinking that later shows up in our work.

There are a few books I use a lot in class because they help me open conversations we need to have. In The Millionaire Fastlane, M. J. DeMarco describes different ways of relating to money [19]. I tell it with the image of a highway: some people travel on the hard shoulder, some move along in the slow lane, and some try to build a fast lane of their own.

On the shoulder, you can spend everything you earn without doing much to prepare for the future; in the slow lane, you can rely on your salary, savings, and a long journey, such as making small contributions to an index fund over years; in the fast lane, you try to build something whose results can grow beyond your hours.

I like the image because it makes you look at where you’re driving your life, what vehicle you have (and how powerful it is), and which lane you’ve chosen, not just how eagerly you press the accelerator.

If all you can sell is one hour of your time, there are only so many hours available. That’s why it’s very difficult for lone wolves, perhaps except for some top executives in large companies, to build true wealth. You can charge more for your hours, and that is a change, but you also need to understand how to increase your income without selling more and more time: the work of other people should yield you some benefit even when you’re not working.

In the next chapter, we’ll look at this through Naval, people, capital, and products. What I don’t want is for you to confuse doing many things with having chosen a model that brings you closer to what you want.

In Polymath, by Peter Hollins, I find another conversation we need to have: how much it makes sense for us to learn outside our speciality [20]. A T-shaped profile combines a vertical bar of deep expertise with a horizontal bar of knowledge that lets you talk with others.

If I know a lot about technology but don’t understand how something is sold, how an investment is financed, or what concerns a people manager, I’ll struggle to follow too many meetings with executive profiles, who are more interested in financial and performance terms.

I can keep widening that shape, develop two specialities like the two vertical bars of the letter π (for example, technology and negotiation or public speaking), or several, like the teeth of a comb (if I also learn personal marketing). I don’t need to know everything; I need to stop treating everything outside my profession as none of my business.

Blitzscaling, by Reid Hoffman and Chris Yeh, helps me explain why a growing company needs to change how it works [21]. Hoffman uses the image of moving from pirate to navy. I explain it this way: you don’t lead a boat the same way you lead a fleet, and you can’t always behave like the pirate captain who was in every battle.

At some point, you need to act more like an admiral: coordinate, choose people to take responsibility, and get different ships moving with a shared purpose. If the company changes size and you keep giving orders in exactly the same way, you may become the problem that keeps it from working. There’s a phrase from that book I loved: “Let things burn.” If you go down to see how every fire is put out or try to put it out yourself, you’ll lose perspective of the fleet.

Solving complex problems has a lot to do with connecting these perspectives. The World Economic Forum’s The Future of Jobs Report 2025 highlights analytical, creative, and systems thinking among the capabilities that matter at work [22]. I’m interested in the practical implication: whenever a problem crosses several departments, an answer from just one speciality falls short.

Moreover, in complex systems, where many related factors are involved, we don’t always know what the real problem is or what its solution would be. Sometimes we can only “dance with the system”: we move something in one place and the problem reappears in another. Think about traffic. You add a lane to a motorway and, seeing less congestion, more people are encouraged to use their cars; the problem can grow.

You install a traffic light to relieve an area and drivers learn a shortcut that collapses another. We have to learn that “dance.” Reading outside your field lets you ask questions that wouldn’t have occurred to you before.

And if I could read only one book in my life, I’d read The Almanack of Naval Ravikant, compiled by Eric Jorgenson [2]. I recommended it to my daughter a week before she started her entrepreneurship degree, as reading to prepare herself and think about what she wanted to build. It brings together ideas about wealth, happiness, learning, and decisions on how to live.

They’ve given me a lot to think about, even if not every page has to fit you. In the next chapter, I’ll pause over some of those ideas and connect them to my experience in business.

As we agreed at the beginning, one person can find each reference and share it. The interesting work comes afterwards: choose an idea from the book and explain how it changes a decision our team makes. The most important thing about a book isn’t finishing it quickly; it’s getting the most out of it by applying its concepts to real life.

There’s a saying that sums up quite well why I like this activity: “If you want to go fast, go alone; if you want to go far, go together” [24]. I want us to keep each other company as we learn, too. When someone discovers one of my posts months later and comments on an idea, knowledge finds me again, like an invisible thread; that person helps me remember something I had already worked on and may not have used for a while.

Also, I’ll confess that publishing increases my reputation: when someone sees all the knowledge I’ve shared together, they’re usually surprised. That person might know as much as I do or more; the difference might be that their knowledge isn’t organised or public.

THE JOURNEY CONTINUES

Oh, I’m sorry.
This is where the sample ends.

The complete book is available now on Leanpub.

You can find more information at www.hardment.com. You can also search for Roberto Canales and “Building High-Performance Teams with Agile Methodologies and Artificial Intelligence”.

I want to keep growing ↗

Check hardment.com for the availability of each edition.

References in this sample

The book’s reference numbers are retained to make them easier to look up.

[1] Bertrand Russell. The Triumph of Stupidity (1933). His reflection on certainty and doubt opens the discussion about how we listen to others. Essay text and Bertrand Russell Research Centre record.

[2] Naval Ravikant and Eric Jorgenson. The Almanack of Naval Ravikant (2020), compiled and edited by Jorgenson and published by Scribe Media. The publisher explains how it was published. I relate Naval’s ideas about wealth, freedom and happiness to my own experience. Wealth can give us freedom and time; what we do with them is another conversation that also runs through his book. He also suggests setting an aspirational hourly rate and delegating tasks that cost less than that rate. Conversation about wealth and happiness, learning to be happy, aspirational hourly rate, specific knowledge, products and media as leverage, judgement and long-term consequences and long-term relationships.

[4] OpenAI. Documentation on data controls and protection. In a personal ChatGPT account, you can turn off “Improve the model for everyone”; new conversations will no longer be used to train models, although they remain in the history. Controls depend on the product, plan and settings. ChatGPT data controls, security and data handling in workspaces and API platform data controls. API rules should not be extrapolated to a personal account.

[17] George E. P. Box. “Robustness in the Strategy of Scientific Model Building”, in Robustness in Statistics, 1979, pp. 201–236. The maxim about models being wrong but useful is Box’s; I remember hearing it from Recuenco and connect it to my aim of building heuristic models for success. Original reference.

[18] Capgemini and Securities and Exchange Commission. World Wealth Report 2026 classifies as an HNWI anyone with at least one million dollars in investable assets, excluding their primary residence and certain personal assets. This differs from the US accredited investor definition, one route to which requires a net worth of more than one million dollars, excluding the primary residence. Capgemini definition, note 1 and SEC criteria.

[19] M. J. DeMarco. The Millionaire Fastlane (Spanish edition cited as La vía rápida del millonario, Sirio, ISBN 9788417399375). I return to the comparison between the hard shoulder, the slow lane and the fast lane from my own reading. Publisher’s details and my review.

[20] Peter Hollins. Polymath: Master Multiple Disciplines, Learn New Skills, Think Flexibly, and Become an Extraordinary Autodidact. PKCS Media, 2020 (Spanish edition cited: Polimatía: Domina múltiples disciplinas, aprende nuevas habilidades, piensa con flexibilidad y conviértete en un extraordinario autodidacta, 2021). I relate this book to T-shaped profiles and the development of several disciplines. Spanish edition and my review.

[21] Reid Hoffman and Chris Yeh. Blitzscaling: The Lightning-Fast Path to Building Massively Valuable Companies. Currency, 2018. A reference for the different stages of growth and changes in leadership. Hoffman develops the transition from pirate to navy in “Uber Needs to Transition from ‘Pirate’ to ‘Navy’”, published on 10 October 2017. Using that image, I explain in my own words the difference between steering a boat and coordinating a fleet. Publisher’s details and my review.

[22] World Economic Forum. The Future of Jobs Report 2025, Chapter 3. Employers’ responses highlight analytical, creative and systems thinking, as well as the ability to tackle complex problems, among the skills relevant to work. Report and survey findings.

[24] Sayings about cooperation and habits. The sayings about going fast alone or going far together, and about action, habit, character and destiny, circulate in different versions; no single author is attributed to them here. Andrew Whitby’s research on the first saying and Quote Investigator’s documentary research on the chain of habits.

[33] Robert T. Kiyosaki. Rich Dad Poor Dad: What the Rich Teach Their Kids About Money That the Poor and Middle Class Do Not!. Plata Publishing, 20th Anniversary Edition, ISBN 978-1-61268-019-6. Kiyosaki popularises the image of the “rat race”; the official PDF includes the section “Escape the Rat Race”. Book on the official site and publisher’s text.

[34] Plaud, Wispr Flow and the logbooks of Roberto Canales Mora. Plaud offers recording and transcription; Wispr Flow offers voice dictation to text. The logbooks are personal projects I describe in Chapter 1; their addresses let readers follow the example through to the resource. Plaud Note, Wispr Flow, robertocanales.com/bitacoras and patriciacanales.com.

[41] Andrej Karpathy. LLM Wiki (2026). A proposal to maintain a linked knowledge base with the help of language models while retaining the original sources. It gives me a working pattern for preserving and connecting knowledge. In this book’s map, we add a decision log and review the relationships and summaries proposed by AI. Author’s document.

[42] Fabrizio Dell’Acqua et al. The Cybernetic Teammate: A Field Experiment on Generative AI Reshaping Teamwork and Expertise. The 2025 work studies innovation challenges completed by professionals at Procter & Gamble, individually or in cross-functional pairs, with and without AI. I discuss the average quality observed during that day: it interests me as a test of a capability, while delivering a product over several months requires other checks. NBER working paper, 2025 and publication in Organization Science, 2026, titled The Cybernetic Teammate: A Field Experiment on Generative AI and Teamwork.

[43] Fabrizio Dell’Acqua et al. Navigating the Jagged Technological Frontier (2023 working paper). An experiment with BCG consultants and GPT-4 on tasks inside and outside the capabilities of the model used. I take from it the need to test specific tasks: the results apply to the model and exercises in that experiment. Our team will have to repeat the checks with its own work and current tools. Harvard Business School research and original paper.

[62] Iván Carrillo and Second Brain. Video about his proposal for organising and connecting knowledge and hardment.com for further information.

[64] Tiago Forte. Building a Second Brain: A Proven Method to Organize Your Digital Life and Unlock Your Creative Potential. Simon Element, 2022. It develops a personal system for retaining, organising and using knowledge outside our memory. Publisher’s details.