In the age of AI, this is more true than ever before. With the latest model of Anthropic Claude Opus 4.7, it apparently tends to do a great job, but only if you communicate effectively at it: if you are precise with what you want and how you want it to be done. If not, there’s a good chance that the model will use more tokens (1.3x its own model card claims) than previous models.

In other words, the worse you communicate, the more effort the large language model has to do, which in turn costs more money. Interestingly, this is very similar to life before AI where you had to communicate with colleagues and if you are not clear about what you’re looking to communicate, your colleagues will have to do a lot of interpretation and they might not do the exact thing that you told them or asked them or otherwise communicated to do.

How to communicate with AI effectively

I’ve said this before, but I cannot recommend enough that you start actively using voice to text. There’s all sorts of software for this. I like to use Aqua, but there are many alternatives that might suit your needs better.

Using software like this allows you to press a button, talk into your microphone, and it will on the spot write whatever you are saying. Even correct little mistakes and do interpretation of the words that you’re using so that you are not having to go back and actively edit text. In fact, the words that you are reading right now are all narrated by me into my microphone and have received no editing whatsoever until now.

So when you are using AI, what you then do is you just talk into your microphone and describe whatever you wanted to do as clear as possible. But you don’t just describe what you wanted to do, you describe very clearly what is the outcome that you hope to achieve. And modern large language models have an enormous context window, so you can really yap on for quite a while and it will not be a problem.

An example

Let’s say that you want to make a website for yourself that shows off your art. AI is really good at this, and you can ask any to make a little website for you right now, and it will do just fine.

Here is what not to do:

AI, please make a website that showcases my art. [you attach your art here]

With this prompt, AI will do just fine. It will create your website with your art. Now, depending on which model and it might ask you some follow-up questions, but this is not a good start. If you don’t tell the painter which color to paint your room, she will choose one. The same goes for AI. It’ll probably be white and boring.

Here is a better example:

I want to make a website that showcases my art. I’ve attached the art, but what I want you to do is make it so that there’s a big screen that has brutalist elements where you showcase two or three pieces of my art at random to a visitor and then allow them to click on any piece and be guided into a gallery of all of my art where we put one in big and they can then click on subsequent pieces of art. It makes those big and gives a bit of an overview.

The website should be opinionated in design, so brutalistic. Make use of a monochromatic color scheme, but if there’s any highlights necessary, use a bright green. The background of the website should be light colored, white or just off-white. And the typography should be equally opinionated.

Please ask me questions about elements of this that aren’t clear or would need further specification

Much more clear to anyone reading what you want, and therefore to AI as well!

I used a ESP32-S3 and some random arcade button I bought online.

One of our engineers responds “My agent made some suggestions, see here <file>”.

I tell my agent to address the suggestions by the other agent.

Yet another agent does an independent review.

My agent also addresses those, and replies to questions.

The (human) engineers merges the new feature.

This all happened this week. Agents and humans working together, but humans mostly acting as glue. And if you squint - the glue isn’t even that complicated to make (in my case, I run an agent in a loop and tell it to check for comments every 30 minutes).

Subsequently, another interesting thing happened, which is that my agent started to reply to comments of my colleagues. But the agent was using my account, so it seemed like I was replying to them. And in turn, they replied directly to my comment, which in reality was written by my agent.

So what I did was I told my agent that going forward, if you post using my account, please say that you are my AI instead of replying as if you were me without any further notes.

It made me realize that there has to be - and yes, I realize I’m not the first person to realize this - there has to be some sort of process or standard by which we say this is an agent doing something. Alternatively, this is an agent acting on behalf of somebody, or this is a human writing the words very directly.
If only to prevent the situation in which somebody says, “Ah, I saw your comment,” to which you have to reply, “What comment?”.

Regardless, we live in an incredibly exciting time. It’s really fun to see new developments happen almost daily and see how they meaningfully can impact our work and our life, arguably very positively.

Until ~last year contributing meaningfully to large codebases by people with zero knowledge of them was still very difficult as AI tools wouldn’t take into account the greater architecture and either under-or-overcomplicate things.

Since the release of gpt-5.4 and opus-4.6, and the fantastic work of the people at Cursor and a bunch of other AI labs, this is now changing. I’ve been able to build a completely new complex feature from scratch in Remote.

After creating a first version that worked, one of our amazing engineers Andrea reviewed my work and left a number of comments. One required a rather large refactor (“this should make use of another part of the app”), which normally would be a massive pain and work.
In my case, I just told Claude to “address the comments by Andrea”, which it did.

A few iterations later in which Claude and I fixed bugs and made improvements, Andrea reviewed the work again ..and merged it!

Almost there

I’m very comfortable with developer tools and version management. That still seems to be a requirement for contributions in larger code bases, but it really is the single remaining barrier.

Apps from scratch anyone can now make with Lovable and others. It’s a matter of time before truly everyone can contribute (which was our vision back in 2013 when we started work on GitLab).

The progress over the past months has been so incredibly sharp that my window for confident predictions about a change in AI model capabilities is at best measured in months: I don’t know how much better AI models will be in one year from now, other than that they will be much better.

We’re not at (or near?) AGI just yet. But programming is ~solved.

Now, this is driven by all sorts of factors. One of them being that we as a society very often feel like you should really be in a place where you can prove that you need a particular subsidy so that and only in that case deserve to get it. And so we put the burden on individuals that really need it to prove that they really need it.

Now there’s all sorts of examples of this, one of them being the TANF cash assistance program in the United States, where there’s all sorts of notorious administrative hurdles, including drug tests, educational requirements, benefit time limits and restrictions on cash usage. In other words, even if you are in a place where you should get this, it’s incredibly hard to get it because the administrative burden is so incredibly big, which further takes away from your ability to do other things with your time, like finding work or taking care of yourself because you are a person that needs to be able to do those things.

There’s a particularly egregious case in the Netherlands in the so-called Toeslagenaffaire, the childcare benefits scandal, where at least 35,000 parents of young children were charged with committing fraud on childcare benefits subsidies and were forced to pay back the subsidies that in fact they actually did deserve to get. This caused an incredible amount of pain. You can read more about it in the aptly called investigation of Amnesty International Xenophobic Machines.

Unfortunately, in my very brief exploration of this topic beyond my cursory curiosity, I found so many examples of this that it would be too much to add here. I will leave it up to you to do further investigation. My point here being is that the people that least could use an incredible amount of work in bureaucracy, unfortunately face the very most of it.

Will AI save us here as well?

As an optimist, I believe that this is one of the greatest positive potential effects of AI there could possibly be. Instead of having to manually fill out forms and having to worry about doing so correctly—there’s examples of forms missing a single number or wrongly written number and causing years of drama— AI can take over this work, not just the filling in of documents. Also, the retrieving the correct information, correcting documents, processing them and generally making it easier for everybody involved and faster. AI can take over this work, not just the filling in of documents, also the retrieving the correct information, correcting documents, processing them and generally making it easier for everybody involved and faster to get benefits they deserve.

I’m not optimistic that societies will suddenly become rational and say “hey, you know what? The fraud is probably significantly lower than the additional cost we have on this massive bureaucratic machine. Let’s make things simpler”

And so I’m putting my hope in technology and hope that for the very most vulnerable people in our societies, AI will also make their life a little bit easier and a little bit better.

Here’s an example. If you want to manage a project today, you would probably start with some sort of project management software. Now, that project management software has been built to support many different kinds of use cases, many different kinds of customers, and it will have a gazillion and one features and decisions made about it that only apply or apply to a very wide group of people.

It’s probably very thoughtful to do that really well for a large group of people. But the reality is, is that if you need something that is just for you or your very specific use case, you could probably build something that is better than its generic solution.

This is where we are today. It’s probably better to build something from scratch that is just for you if only you are going to use it. And the reasons for it are actually manyfold. First of all, it will solve your problem better than any alternative.
But second, it will also be free. You will not need to pay somebody money. The only thing you have to pay is essentially the tokens of an AI provider, and those are not very expensive. And so you will have something perfectly made for you, and it will be free.

What about big software?

An argument that people running big software companies make is that this is not going to happen to them because their software is big and complicated and does a lot of things for really large organizations.

I’m actually not sure whether that’s true or not. There’s two arguments you could make.
The first one is that big software is really big and the cost of replicating it is going to be far greater, even just in terms of time and effort with AI, than it is to just buy it off the shelf. This might actually be true because the software is not just selling the software, it’s selling transformation and it’s selling solutions to problems.
Now, the counter argument is that any organization which is willing to spend hundreds of thousands, if not millions of dollars on software can probably free up sufficient number of people to full-time work on big software just for them. And that might be sufficient.
Now, as a last counter to the above, you could argue that big software companies will transform themselves as well. And they’ll just get proportionally more ambitious and proportionally better at solving the problems that they set out to solve.

And so maybe even if you do dedicate a small number of people to build software custom for you and your organization, it’s not going to be worth it because that company that is selling software just does a better job regardless, no matter if everybody has amazing AIs available to them.

I don’t know what the future for software is going to be, but for sure it’s going to be interesting.

And so at 30 weeks, in the very middle of pandemic, while Remote was taking off, our son was born. He was smaller and more fragile than our daughter. As the pandemic was in full force and no vaccines available, I was only allowed to see him a week after he was born. Six weeks later was the first time I was able to hold him at all. He stayed in the NICU for almost two months.

Today, both kids are happy and healthy and have little to nothing to show for the fact they were born prematurely. We were lucky with their outcomes, because it’s not always like that. In fact, premature births is the single largest cause of death of infants before age five.

To make the life of parents a little bit more informed, and maybe make them feel a little less alone in the journey of having a premature kid, Carla wrote a book about both our experience and science behind those kids that spent too little time in the oven.

That book is Born Strong.

Get the book

You can buy the ~300 page book today at bornstrongbook.com.

All proceeds of the book are going to organizations doing research into prematurity and supporting parents of premature kids. We are donating copies to NICUs around the world.

The English version is available as paperback and digitally. The Portuguese and Dutch versions are set to arrive in a month or so.

me: ”You can even build it yourself, just throw it in Claude!”

Pedro: “No I did. I built it. That doc just explains how it’s structured.”

Never wrote a line of code in his life. Build a very complicated, very useful app that you can’t buy off the shelf. From scratch. No vibe-code app. Just Claude Code / Codex.

Until very recently it was entirely possible to build a little app that did a little thing. But large projects required many iterations.
Not anymore.

If you give the AI enough to work with, are specific enough with what you hope to achieve, it can now single-shot a genuinely elaborate and complex application.

We’re beyond the threshold

As I keep reminding you: today AI is at its worst. So before you comment “but you still need someone to review code” or “you still need programming skills for larger projects” - yes you’re entirely right. But that’s today, the worst day of AI forever going forward.

At Remote, already 70% of all code is written by AI (that we can track), bugs are auto-fixed by AI (when possible) and code reviews are given a first pass by AI. This was meaningfully different three frikkin’ months ago.

The path from 70% to 99% is very clear, and I don’t see any reason that we’re soon at 100%. You no longer need to learn a programming language. Rather, you need to become great at translating your ideas to words to make great applications.

Progress is moving so fast that people and companies that won’t adopt this new way of working will be completely obliterated by people that do, as it now takes hours what used to take months.

The onus is on you. It’s not a requirement anymore to have a particular education or experience in tech. If you can talk, you can build. It might be slightly bumpy today, but it’s getting easier every single day.

Don’t know where to start? Ask you favorite AI.

I have an agreement with myself. The agreement is that things that I do for fun, I want to keep fun. And I found that once you, and I’m not the first one to realize this, once you make something fun mandatory or more like a job, it’s a really easy way to make it not fun.

As an adult, it’s very hard to give yourself the space to do things just for fun. But I found that doing things for fun is one of the best ways to stay creative, to explore new things, to explore something that you’ve never done before, whether that’s learning a new skill, doing something artistic or creative or something else entirely.

Play as an adult is difficult. There are so many expectations from society, other people in our professional environments, that we’re always busy improving ourselves / leveling up. Having pressure to achieve or perform is not conducive to creativity. Learning something new is tiring and requires you to overcome constant hurdles. If the subtext of your learning is that ultimately this doesn’t provide you with additional skills that help you in your professional lives, it’s much harder to do so.

However, if you give yourself time and space to just play, to enjoy something, whether that is learning a new skill or doing something you’ve never done before without the burden of you becoming a better professional, i.e. by having zero pressure projects, you can explore new parts of yourself and enjoy things for what they are.

We are quickly approaching a time where being a professional, especially if you’re working in an environment of white-collar workers, is less about having a particular degree and more about being creative with the tools at hand. Those tools, of course, being AI. I would argue that play, exploration and ultimately curiosity are some of the most important new skills to attain.

Today AI is at its worst that it’s ever going to be going forward. But it’s very easy to underestimate what that really means. A good mental exercise is to imagine not ten AI agents, but for example, ten million.

I urge you to imagine is that we will actually get to the point where you have an equivalent of 10 million AI agents, and that’s not decades away—that is mere years away from today.

It’s very hard to imagine what it means to have 10 million employees that do exactly what you want and essentially never get tired. But that’s also the point of the exercise. AI is going to get 10 million times better than what it is today. What is going to happen in the world and what is creation going to feel like when at any moment, at any time you have ten people equivalent intelligence available to you to do whatever you want them to do?

Obviously, I don’t have an answer, and I don’t think anybody has a reasonable answer to this question.

I have a lot of hope. I’m an optimist, and so I like to think that what are problems today, many of those will disappear. At the same time, it’s easy to imagine that there’s people that use this kind of power for nevarious purposes.

Whichever way the future is leaning, I’d like to challenge you to think about the 10 million AIs that you will have at hand. From now, where you have 10 to the 10 million agents, what are you going to do to make sure you’re ready for the future? And how are you going to influence that future to be better than what the world looks like today?

Share

Nice notebooks. Leichtturm1917 makes great notebooks. Because I use fountain pens I use their thicker-paper version (Square).

Subscribe now

Vestaboard. Okay, so this is really nice and really expensive. I’ve wanted one for years and now finally have the large black version in my office. You don’t need this. But if you do get one, it’s super cool. I display random things on it using an app I made myself.

Fountain pens! This is a world where you can dive deep and not surface for a while. Muji has a great fine-tipped fountain pen, but if you can try to get a fountain pen with a plunger to refill it. This saves you from refilling it with disposable cartridges, and allows you to dive further into the world of inks. If you’re willing to spend, check out the amazing Lamy 2000.

A mechanical keyboard. I need to make a whole post on it (I have one on draft, and it’ll be long). But if you type a lot, it’s worth making that feel nice. Keychron makes great keyboards at reasonable prices! I would recommend against one with slim keys, but that’s a personal preference.
Oh wait, I just realized that this is connected to your computer. Well, it’s nice anyway, and I’m the boss of this post.

A non-plastic water bottle. Kleankanteen sells nice stainless steel water bottles. I’ve used them for many years and they are nearly indestructable.

A CO2 monitor. High CO2 leads to lower cognitive performance. If you’re in a closed space, CO2 will quickly increase (because of all that breathing you do). Get a good CO2 monitor like the Aranet4 to know when it’s time to open a window.

What’s your favorite desk item?

Below on the apps that are still running. I made 3-4x more that are simply not live anymore or not used anymore.

Central app monitoring: An app that tracks that status of all other apps. My deploy SKILL.md automatically adds new apps to this one.

Running streak overview: My wife runs every. single. day. This app takes data from Strava and plots it in fun ways over time, including upcoming races and whatnot.

Rekensterren: A fun mobile app for my kids to learn to do math, and earn Pokemon cards when they reach certain achievements

Leessterren: The same but for reading for my five year old

HIIT Record: An app to show our trainer which HIIT workout was the best/worst based on Strava data (heart rate mostly)

Vestaboard: I have a Vestboard (google it, it’s an amazing thing). This app controls it and shows me things I’m interested in, and works well with the rest of my apps, e.g. showing when things are down.

MQTT-tracker: My home automation runs on MQTT, this app allows me to see all MQTT messages come by and plots them over time. It also has a function that finds correlenation between particular signals. Very fancy.

Heat pump programmer: Long story short: the thing that controls my heatpump doesn’t work well. I made my own controller that takes as input temperature and valve openings of the floor heating and controls the heatpump based on that. It’s elaborate and works quite well - much better than the software that the supplier came up with.

Ice cream app: I got a Ninja Creami, and this app suggests recipes and stores my favorites!

Piano sight reading: My daughter and I are learning to play piano. To improve our ability to know which note is which key, I made an app! Has achievements, built-in spaced repetition and a nice onscreen keyboard. In theory it can connect with MIDI to our keyboard, but I learned the hard way that Safari on iOS does not support webMIDI.

Guitar tabs: An app to keep track of guitar tabs I like. I can import any and make them easy to read and print.

For work, I made a million others. That another day. What’s your best creation so far?

Here is how it works: Below you will find a list of terms and concepts to learn about. I’ve given you prompts to get started, but you can also write your own. Use your AI of choice to learn what they mean and how to do things. If you think you got it, move to the next item. By the very end, you will have taught yourself how to make a simple web app that runs on your computer using Claude Code in the terminal.
I strongly recommend using a Mac, but if you must insist in using inferior systems, you can get AI to help you run this on a Windows machine as well.

If you already know about something, skip that!

Why Claude Code?

Programmers at the company that you’re working at today are writing code and directly committing that code to a greater code base. Using Claude Code, you will be able to do the same thing.

So even though you might today be able to make apps using something like Lovable, you cannot actually modify the code that your company is making, and this will allow you to do that. So it’s useful to do this.

Now, there’s a big downside to using Claude Code for little vibe-coded apps, which is that you have to figure out deployment—as in, where do you put the code so that it keeps running other than your own personal machine. But for today, I’m just going to give you a taste of Claude Code rather than teaching you how to deploy and manage an app in the cloud. That’s okay. You don’t need to learn that today.

Instead of Claude Code, you can also use Codex CLI by OpenAI. Everything will work the same way.

Cooking with Claude Code

Okay, we are going to make a little application that runs on your computer where you can enter what you have in your fridge and then it will give you recipes.

1. Web app. Teach me what a web application is, what frontend means, backend means, what the database does and how these pieces connect. Also teach me the concepts of running something locally vs. “in production” or “in the cloud”.

2. Terminal. Teach me what the “terminal” or commandline. What is it for and can you give me step-by-step instructions that make me create a folder somewhere on my computer? Make sure I have the tools necessary to easily navigate in the terminal.

3. Install Claude Code: https://claude.com/product/claude-code and create an account (if you already have one, use that!)

4. Now create a directory in your home folder and call it something fun

5. Run Claude claude

6. Now tell Claude what you want to build! Here’s a nice prompt that will get you something cool:

   1. Build me a web app that allows me to choose ingredients that I have in my fridge/house and then suggest recipes based on that. Remember what ingredients I have in house by storing that in a database. Make sure to see the database with a lot of different recipes and make it possible for me to add new recipes with ingredients. Also, create a Makefile that makes it easy to run this app locally by just typing make run. Make a plan first before starting to work on this. Then use sub-agents to work in parallel on different features at the same time and start work.

7. Let Claude cook for a while, make sure to allow it make to changes while it’s doing it. You can set permissions to always accept by running claude with claude —dangerously-skip-permissions, but the name says it all. Run that with caution.

8. When Claude is done, you should be able to run make run in your terminal and see your app in your browser (probably by going to localhost:8000). If you’re stuck, you can just ask Claude for help!

Next steps

Congrats, you built your first app using Claude Code!

What you can teach yourself next:

• Versioning, and specifically Git. This is a toughy to crack, but crucial for contributing code! Ask your favorite AI this to learn it:

  • I know nothing about versioning or Git. Please tell me the fundamentals and teach me how to use Git day-to-day and how to, for example, clone an open source project and how to make contributions to something like that.

• Deployment: how do you actually get your app in front of the world. Here’s a good one you can run in the folder you just used:

  • Now, how would I go about deploying this? What services are available to me to be able to make this available to other people on the public internet?

But alas, that time has come to an end. If you are able to write any words, you are now able to conjure the spirits of programming right on your very computer.

Designers, they knew about the secrets of margins, padding, and typography. Unless you were born with magical eyes that are able to perceive whether something is good and whether something isn’t good design, you would forever be lost in what can only be called shitty UX.

But alas, the time has come to an end. If you are able to write some words, and specifically the words “make it pretty”. Then you’re able to make something that passes as a reasonably good design.

At the end, there are the product managers, both the highest paid and least well-defined occupation in the Triforce of application builders. What do they do? Nobody knows, but also their time has come to an end. For now, you can simply speak the magical words: “Build a bunch of features that make sense”.

The time of specialists is coming to an end, and we are entering the age of builders. We are all builders, some of us with better taste than others. Some of us do certain things more than others, like thinking about how something is built rather than how something looks. But at the end, there are only builders. No more engineers, no more designers, no more product managers. Just builders.

Subscribe now

This is something that I’ve been seeing: the larger the organization is, the less likely individuals are directly contacting the CEO, for whatever reaon. This is incredibly bad for the organization because the CEO needs to be able to act on what people in the field/on the ground are seeing, feeling and hearing. Unless you actually directly tell them, the only way the CEO is going to hear about what is happening on the ground is filtered through layers of management. In and of itself, not a bad thing because indeed not everything is priority. But if you feel something is important and you work at that company for a while, you are probably right that it is important.

Similarly, if you are a customer of a given business, the same dynamics happen. I sometimes get emails from customers that apologize for contacting me or see reaching out to me as a last-ditch effort in resolving some conflict. Whereas in reality, I prefer to be rolled into conflicts much earlier because I have really high leverage in solving those.

Subscribe now

All of this to say that I feel quite strongly that you should actually DM the CEO if you have something important to share with them, whether you are a customer or whether you’re working for that company. Maybe your manager or even your manager’s manager might not enjoy this because they feel like you’re bypassing them. But in reality, if you feel that something is so important that the CEO should know about this and it is so urgent that you should definitely DM them.
What is important is that you don’t doubt yourself. It doesn’t have to be the end of the world. Really, anything related to customers is probably on top of mind for whomever’s leading the company, and so therefore it might be worth sending a message.
On top of that, yes, they might get a lot of messages every single day, but sending a short message is not a big bother to anybody, or really it shouldn’t be.

Whenever I have really strong feedback, be that positive or negative, I tend to try to do this. I send a message to the CEO of said company of which I’m a customer, and 50% of the cases I actually get a reply. And there’s no correlation between the size of the company and the reply. And I’m pretty sure that in the other 50% of the cases, they still see my message and they might do something or not with it. But at least I did an attempt and it didn’t hurt anybody to do so.

Likewise, it will not hurt you to send that message. If you have something to say, send a message to the CEO. It’s okay. The world is not going to end.1

Share

If you must ask this question, the answer is no.

OpenClaw is a super powerful way of giving AI access to your machine and messaging services. It can do things we all dreamt of seeing AI do. Think “Her” (highly recommend watching that movie now!). It can order groceries for you, build apps for your needs, but also answer phone calls and send messages to friends.

It’s so effective, because you can truly give it the keys to the kingdom - every single piece of your data and access. You don’t have to, and you can run it safely, but if you’re asking whether you should you’re likely not in the camp of having a deep interested in figuring out how to limit its powers.

It’s very new and it is getting an enormous amount of attention (and about $50mm in investment as of last night, I believe). That will most certainly go to making it easier and safer to run OpenClaw. Plus, it’s technically not a very complex thing, so there will be dozens of copycats.

Give it some time for the tech to mature and then try it on a computer where you don’t have personal credentials or data.

Or - if you’re like me - go through great lengths to run it on a machine that you carefully lock down and restrict. It’s a taste of the future.

Subscribe now

But with the rise of AI, we now also have tools like Lovable and Claude Code that allow you to actually build solutions for your problems.

And I think this is where the world is going. We’re going to be buying less and less off-the-shelf software to solve problems that we have, and more and more build one-off solutions for ourselves.

Sometimes there will be things that we build that we will continue to use for a very long time, but I also think we’re going to be building throwaway software—software that solves a single problem for a short period of time and then you get rid of it. Then later you build something new that is more aligned with what you need at that very moment.

Getting your hands dirty

Last night when I told a colleague that this is where I think the future is going, they told me “well, I don’t think I have that many problems to solve”. Which sounds like a really good position to be in, but maybe you’re simply underestimating the power of AI.

What I want to propose you to do is to actually start building something today. The best way to start building something is to play, to try out things. You don’t have to solve any big problems.

Maybe the problem that you’re trying to solve is that you want to automatically buy new coffee when it runs out, or maybe you just want to read the news in a format that perfectly suits you, or maybe you just want to keep track of your runs.

Or maybe you want to create a database of things that you have in your closets. There are many different things to build, and what I suggest you do is the following.

1. Choose an arbitrary problem. My suggestion is to create a simple database of sorts. For example tracking the food you have in house.

2. Sign up for Lovable

3. Try to build it!

It’s really that simple. What you should try to do is to get to a place where either you are unhappy, and then rather than iterating from that point, you just start fresh with what you’ve learned about how Lovable works, about how building with AI works. Or alternatively, you get to a place where you’re actually happy and then try to use the app. Try to make it useful for you. What are the decisions that you have to take to make it useful for you? How can you make it easier? How can you make it better? And maybe what you’ll realize is that you’re done and you solved the real problem for yourself. Or what often happens to me is that you realize, well, this app really didn’t need to exist, and you just tried to solve another problem.

The point isn’t that you solve a real problem right now. The point is that you get comfortable building something from scratch with these new tools.

The more practice you put in, the more you learn about how these systems work and the more you learn to solve problems differently.

And that is the skill you’ll have to develop in the world of tomorrow.

Start building!

Share

I don’t want to fill a form. I don’t want to learn how to dismiss your onboarding use your app. I don’t want to create a new entity. I don’t want to edit my profile. I don’t want to setup permissions. I don’t want to navigate to a different page, to then navigate to a different tab. I don’t want to configure notifications. I don’t want to use the app at all.

Don’t make me do anything. Just solve my problem.

Ask me the very minimum of what you could possibly need. Make it ridiculously easy to give it to you and then don’t bother me. Solve my problem and make it disappear forever.

If there is something I should know, tell me. Otherwise don’t bother me.

Make it so I can just tell you what I want. Don’t make think about how to ask. Don’t make me find an email, or go through a funnel to tell you what I want.

Give me a text box, let me talk/type what I want and then magically make that happen. But only if my ask is not standard. Because, remember, I didn’t want to do this at all. I just wanted my problem to be solved.

Don’t make me guess. Don’t make assumptions that I know what you know. I don’t know anything. I just want my problem solved with the minimal effort on my part so I can spend more time playing video games delivering shareholder value.

Share

But truth be told, Googling is a skill. You need to figure out how to search for things and be able to evaluate the validity of the contents.

I was brought up with the idea that you couldn’t use Wikipedia as a source. And so, there was still value in education, in the raw materials that education would bring to you because they were carefully put together. They helped you build knowledge upon knowledge upon knowledge so that you become somebody that knows something.

But that time has passed. You don’t need to learn to Google anything anymore. You can now ask an AI for anything, and you will get a mostly, if not entirely, accurate answer. More than ever before, you’re now able to ask clarifying questions until you understand.

This is one of the greatest changes in learning and education that has ever happened. Up until now, if you wanted to learn, let’s say, about neuroscience, there was a point at which you would either have to dig into books and books and books to figure to form an intuitive understanding of particular systems in the brain. You would have to find a teacher that could answer your questions no matter how silly or complex, just because the way the information was presented in books didn’t necessarily match how your brain organized the information.
But now you have the world’s greatest teacher at hand at all times. You can ask an AI to teach you about neuroscience, and then you can ask it to teach you in a way that suits you, and you can keep asking questions.

AI is endlessly patient. It will keep answering your questions for as long as you keep asking them, and with each passing month, the AI is getting smarter and better at doing this as well.

Today you have the world’s best, most patient teacher available to you at any point in time. And these are teachers that know at least the basics of almost anything and can teach them to almost anybody at any time.

Everything always at hand

Thus, the time to have an excuse of not knowing something has passed. Sure, you might not know something. In fact, you’ll probably not know lots of things. But teaching yourself something new has become easier than ever before.

Certainly, some things are hard to understand and will take you lots of time and study, even with AI, to truly grok. But access to that is available to everybody.

My personal pet peeve is when people call themselves non-technical. Essentially what you’re saying is that, hey, I’m good at lots of things, I just can’t program.
That time has passed too! If you want to learn to program—and really, I shouldn’t have to tell you this—you already know what to do. You simply ask AI “hey, teach me how to program. Where do I get started?” You can keep asking clarifying questions until the very point that you deeply understand it. And you can go as deep as you want.

Now the burden of knowing is on you and nothing else.

Experience and depth still count

Now there are some limitations to this. The first one being experience. There’s still inherent value to experience. Not all human experience is encoded in the AI, and so AI is not be able to share that necessarily with you. At the same time, it’s not possible for you to learn everything all at once and instantly grok all of it. You will still have to spend time, flexing your brain.

Yes, you will still need to study, and studying takes time, and learning things takes time. But, for the betterment of your life or work, there are not a lot of excuses anymore to not know something.
Companies are now changing faster than ever before. It’s up to you to change along and gain the skills and knowledge to keep up.

If you make time for learning, which I realize is a big if, then there really shouldn’t be a reason for you to not become technical, to not grow along with the company and have the knowledge to thrive in this day and age.

Share