One of the most depressing moments in life is looking around your room and seeing all the stuff you bought with your hard-earned money that you no longer use.

Money that, if you’d put it into Bitcoin or NVIDIA stocks, would have… well… let you buy even more things.

You know what….I sense a pattern here.

For me, I have this whole workout kit I bought when I decided I was finally going to get fit. The hoards of books I bought because I was going to become one of those people who reads a book every week (I do try, though). And the gadgets I convinced myself would help me be more productive.

It’s a little embarrassing how much money I’ve spent trying to become a better version of myself without actually doing the things that would make me better.

And I’m 100% sure that 99% of you reading this have been through the same thing. Most of us have a drawer, a cupboard, or some forgotten corner of the house dedicated to things we were absolutely convinced we needed at the time. We bought them with the best intentions… I’m sure we did use them for a while, but eventually we moved on to the next thing that was going to change our lives.

But this got me thinking. Why do we want so much stuff in the first place? And why, even after we get what we want, does it rarely feel like enough?

Ok, so one possible reason is that we, as a generation, might have the perfect combination of gullibility and impulsiveness to help any business thrive (Just ask the guys at Apple!). Or maybe there’s something a little more interesting going on here…something that goes beyond…you know…flashy advertising or questionable financial decisions. Maybe there’s a more human reason as to why we have always been soooo good at wanting more things.

We spent most of our existence trying to get more.

For roughly 96% of our species’ existence on this planet, humans lived as hunter-gatherers. Agriculture only emerged around 12,000 years ago, which means that the “Modern” societies we live, love and thrive in make up only a very tiny fraction of our combined evolutionary history.

For hundreds of thousands of years, our ancestors survived by finding food, securing shelter, making tools, and figuring out how to get through the next difficult season.

And in those days, if you found a reliable source of food, you didn’t congratulate yourself on achieving work-life balance and post about it on your cave wall and move on. You made sure you could keep getting it.

Useful tools were also worth holding on to, and so was any opportunity to secure resources that might save your family or tribe from starvation.

For our ancestors, having more probably meant having a better chance of surviving whatever came next.

There was a social side to it, too. We lived in groups, and our standing within those groups mattered. Being useful, respected, having useful relationships, and having access to resources could all affect our chances in life.

Evolution didn’t need to make us perpetually dissatisfied with things. It just needed to favour people who were really good at spotting opportunities and making the most of them.

Our ancestors spent their lives trying to get their hands on whatever resources they could. We’ve come a long way since then, but I’m not sure we’ve quite figured out when to stop. We don’t even have to hunt for our food anymore. We can just Swiggy or DoorDash it and still spend hours researching which air fryer to buy.

We’ve basically solved the problem of finding food and moved on to the more pressing matters… like our kitchen aesthetics.

Progress, I suppose.

I mean, we evolved in a world where getting more things could improve our chances of survival. Now, it’s almost a hobby.

You get a better salary, and after a while, it becomes your normal salary. You finally buy the car you’ve wanted for years, only to start noticing the nicer ones around you. Even your new phone starts feeling old once the next model arrives. (Just ask the guys who use Apple!)

I mean, there is something almost comical about an animal that evolved to seek out resources for survival now having to watch videos on minimalism and make it their entire personality ( I am looking at you, Matt D’Avella). We’ve changed our surroundings beyond recognition, but I think some of our instincts are still very much alive and well.

Surely someone has tried to fix this?

Of course they have.

For centuries, religions and philosophical wings and Mastercard(remember those ads?) have encouraged people to practise restraint and questioned whether wealth is a reliable path to happiness. Certain Monastic communities have gone as far as to live their entire lives around giving up material possessions.

And then there was communism.

When Karl Marx published Das Kapital, he wasn’t trying to teach people how to declutter. He was trying to figure out why a capitalist society was so obsessed with accumulating more and more wealth. And I don’t mean people buying too many things. I mean the entire economic system seemed to be built around the idea that capital should keep growing, pretty much forever.

Marx had a problem with how this worked. The people who owned factories and businesses could keep getting richer from the work other people did, while those workers had to sell their labour just to make a living. His solution was to have the means of production owned collectively, instead of leaving them in the hands of a small group of people.

I’m not going to pretend I can explain all of Marx in a few paragraphs. But there are a few interesting ideas there. We usually tend to think of wanting more as a personal trait. We sure can argue that some people are more ambitious or materialistic or just really bad at resisting a well-pitched sale. But what if the system we’re living in also had something to do with it,?

After all, the economy doesn’t just decide who gets how much money. It also influences what we consider success, the quality of things around us, what we spend our time chasing, and how much we feel we need to have before we can finally relax.

If a system rewards accumulation, people have good reason to… well…accumulate. Change those incentives, and you might change the way people behave…right?

Well, you see, the problem here is that changing who owns things doesn’t automatically change why people want them. These ideas, as I see it, never account for the basic human nature of simply wanting more things….

I mean, the Soviet Union tried to implement communist ideologies and got rid of private ownership of many major industries, and Mao’s China tried to take things even further, trying to change almost every aspect of society. Both experiments came with horrific human costs, including catastrophic famines. Obviously, there’s a lot more to what happened in these countries than just people being greedy or materialistic.

But one thing worth thinking about is that getting rid of private capital didn’t magically get rid of people’s desire to have or want more things. People still wanted power, status, and access to things that others couldn’t get. Turns out, even if you change the rules around money, people are still pretty good at finding other things to compete over.

Now, I’m not saying this proves that building a fairer society is impossible. The way we organise an economy does matter, and who controls the resources can make a massive difference to people’s lives.

But there’s a difference between changing a system that encourages people to accumulate wealth and somehow getting people to stop wanting more in the first place. The first is a political and economic problem. The second is something we’ve been trying to figure out for a very, very long time.

We invented a world that feeds the instinct.

Now, capitalism is another interesting take on the same problem. For all its flaws, it has done a pretty incredible job of turning our desire and want for more into an actual economy.

You want something? There’s probably a company making it. You want a better version of it? There’s probably another company working on that (Just ask you know who). And if neither exists, give it a few years. Someone will eventually come up with it.

And to be fair, this has given us some pretty amazing things. Medicine, technology, the internet, Airbnb. All the stuff that would have sounded like science fiction to someone living a few generations ago(Especially Airbnb). The markets and competition have played a big part in getting us here.

But there’s a funny little side effect to all this. When you get really good at giving people what they want, you also get really good at finding new things for them to want.

Think about how many products are sold to us with the promise of becoming a better version of ourselves.

And for a brief moment, buying the thing feels a little like becoming that person…but that is just a phase, and it fizzles out. But hey…there is always something else to buy if the first thing doesn’t work out.

And I don’t think there’s some grand conspiracy behind this. Companies want to make money. They make products, find people who might want them, and try to convince those people to buy. That’s how the whole thing works.

It’s just that we’re particularly easy to sell to when the product promises to make us a better version of ourselves.

Combine that with our tendency to always want a little more, and you end up with a world where there’s practically no end to the things you can buy to improve your life.

The slightly unfortunate part is that you still have to do the improving yourself.

The problem with having enough

I don’t think the answer is to stop wanting things.

Wanting more isn’t necessarily a bad thing. It’s what gets people to build businesses, learn new skills, and try to make their lives better. I don’t think there’s anything wrong with wanting to do well for yourself.

But somewhere along the way, it’s easy to start confusing the things we want with the things we’re supposed to want.

Social media makes it a whole lot worse. You open Instagram, and there’s someone buying a house in London, someone travelling through Japan, and someone else who has somehow become a millionaire at 23 and some random bald guy calling you a beta male for not buying his courses.

And the annoying thing is, you might have been perfectly happy before opening the app.

Maybe you weren’t even interested in buying a house in London. But now you’ve seen someone else do it, and suddenly your own perfect house in the suburbs of Kochi feels a little less impressive. You simply start picking up other people’s ambitions without really noticing. Things you never cared about become things you feel like you should care about.

Of course, there’s nothing wrong with wanting a house, travelling the world, or making a lot of money. Money gives you options, and having a comfortable life is hardly something to feel guilty about.

But I do wonder how often we stop to ask ourselves what we actually want, as opposed to what we’ve been conditioned to want.

I can imagine having plenty of money, a nice house, excellent gadgets, and an impressive collection of books. I can also imagine spending so much time acquiring all of them that I never get around to reading the books, enjoying the house, or doing anything particularly interesting with the money.

I mean, think about spending all this time collecting the things for a good life, only to run out of time to actually live it.

And that brings me back to the dumbbells.

I bought them because I wanted to get fit. They’re still there, doing a fantastic job of taking up space.

I could blame capitalism, my “hunter-gatherer instincts”, or Steve Jobs. I could even write a lengthy philosophical essay about why humans are so obsessed with accumulating things.(wink..wink)

Or I could pick up the dumbbells.

I think the last option would be considerably more effective.

Still, if NVIDIA keeps going up, I might be able to get a nicer pair…one with proper grips and all.

There is nothing quite as powerful as a power cut in your area to make you disconnect from the online world and ruminate on life and all its wonderful possibilities.

So ya… here goes.

At one point in every person’s life, they consider becoming a farmer.

I’m not sure when this happens exactly. It might be after a particularly bad day at work, or when you’re sitting through a boring meeting. Or it happens on a random weekday afternoon when you look at your screen, close your laptop, stare into the distance, and think, You know what? I could just grow some potatoes.

And suddenly, the idea makes a whole lot of sense.

Like, you start imagining a little house somewhere in the quieter part of your state or country. A few acres of land, maybe some chickens or a dog. You wake up early, grow your own food, and spend your days doing some honest work. No emails, no deadlines, no one asking you to jump on a quick call or anything.

Just you, the plants, and the sound of crickets.

Of course, you have absolutely no idea how farming works. The closest you’ve come to agriculture is keeping a plant barely alive for two weeks. But that’s a problem for the future version of you. The current version is busy browsing properties in Munnar or Wayanad. You can figure out the farming part once you’ve bought the land.

For most of us, this fantasy lasts for about twenty minutes. Then the rent is due, a few bills show up, and you remember that you wouldn’t know what to do if someone handed you a cow or a goat.

Lately, though, I’ve noticed that this whole farming thing seems to be coming up a lot more often, particularly among people in tech.

And I think we can blame AI for at least some of it.

I mean, every other day, there’s a new announcement about a model doing something that would’ve taken a team of engineers a week not too long ago. The tools keep getting better, the industry keeps changing, and everyone is trying to work out what this means for their jobs. Somewhere in the middle of all this, a developer decides he’s had enough and announces that he’s going to buy some land and grow vegetables.

Honestly, fair enough.

But there’s a certain irony to it. I mean, we spent years learning to build software, automate things, and make computers do more of the work for us. We moved everything onto screens and wholeheartedly celebrated and adopted every new tool that made our work feel faster and easier.

Now that the tools are actually getting really good at the work, some of us are looking at a patch of land and thinking how growing some Vaazha(banana plant) might be a safer long-term investment.

I mean, at least you can eat some pazham(banana) .

I’ve been thinking about why this idea is so appealing. I think there’s something satisfying about the kind of work where you can see what you’ve accomplished at the end of the day. You plant a seed, look after it for a while, and eventually you get a tomato or a carrot. You can hold it, eat it, sell it, make something with it. You don’t have to put together a pitch deck explaining why the tomato matters, and nobody’s going to ask you about your customer acquisition costs.

Comparing that with building software today….you can spend weeks working on something, finally get it into a state you’re happy with, and then watch a new AI tool reproduce most of it. You barely have time to enjoy what you’ve made before you’re back to wondering what you should learn or iterate next.

It’s a strange way to work. You’re always moving, always catching up, and somehow there’s still a feeling that you’re falling behind.

Actual farming, of course, is a whole different story. Farmers deal with unpredictable weather, pests, rising costs, and of course, the hard and enduring physical work. The soil doesn’t care that you’re having a difficult week, and there is no detox or sabbatical that you can take from farming. And I’m fairly sure that the whole romance would wear off once you spent an entire day in the rain trying to save a crop from some disease you can’t pronounce.

But the fact that so many of us keep on entertaining the idea is in and itself quite interesting.

I think we might be just tired of feeling like we have to keep up with everything. You learn a new skill, and the tools change. You get comfortable with a way of working, and there’s a new way that’s supposedly ten times better. You spend years getting good at something, and then OpenAI releases a new version that does the same thing in minutes.

Even when things are going well, there’s always something new to worry about.

I suppose farming offers a welcome change of pace.I mean, A mango tree isn’t going to produce fruit any faster because you’ve been particularly productive this week. Things happen when they happen, and you have to learn to live with that. There is something quite nice about the idea of working at a pace you can’t optimise your way out of.

Of course, give a developer six months on a farm, and they’ll probably build an automated irrigation system, install some sensors in the soil, and spend the evenings building models to predict the next harvest. Of course, we’ll somehow find ways to turn farming into a dashboard eventually. I mean, we really can’t help ourselves with that.

Still, the thought makes me wonder what we’ll do if AI really gets super good and it really does give us more time and make us capable of doing much more with less effort.

Knowing us, we’ll probably find something else to fill it with. Another project, another skill, another thing we can optimise. I wonder if we’ll ever get around to just enjoying the extra time.

I’ve always liked building things. I like the feeling of starting with an idea and slowly turning it into something real. But sometimes I wonder whether we’ve become so focused on making things faster that we’ve forgotten to enjoy making them at all. I mean, there’s a difference between getting something done and enjoying the process of doing it. And somewhere between all the noise, that distinction can get a little blurry.

I guess that’s why the farming fantasy keeps coming back. It gives us that permission to imagine a life where we aren’t constantly measuring ourselves against the next thing. You know… where we can spend a day doing something and feel reasonably happy with what we’ve achieved.

Or maybe I’m reading too much into a bunch of tech workers joking about buying farmland. That’s also entirely possible.

Anyway, I say all this as someone who would probably have to Google the tree that tomatoes grow in.

So, clearly, I have some things to work on before I become a farmer.

Still, I’m keeping farming as an option.

Just in case.

Alright, so nobody wakes up on a Tuesday morning and thinks, you know what this world needs right now?……another poem.

I mean, who in their right mind would open their notebook, look at the state of humanity, and conclude that what we’re really missing is a four-line rhyming verse about a dying flower?

And yet, someone writes it.

The same goes for products. Nobody wakes up thinking, you know what’s missing from my life? Another app. Another way to take photos of my food and share them with everyone so they know I have a life.

And yet, someone builds it.

The poet typically sits alone somewhere trying to put feelings into words that have probably existed for thousands of years. The product builder sits in front of a screen, or several, trying to solve a problem that people might not even have figured out how to describe yet. Both end up spending an unreasonable amount of time on something that, to everyone else, seems entirely optional.

And the funny thing is, neither can quite explain why they feel the thing needs to exist. They just have this idea they can’t seem to leave alone.

Occasionally, that turns out to be enough.

The audacity of making something nobody asked for

There’s something a little weird about creating things, especially when you think about how much time people spend doing it without any guarantee that anyone else will care.

Before Shakespeare wrote Romeo and Juliet, nobody was submitting a feature request for a tragedy about two teenagers with shitty family dynamics. I mean, I doubt there was a petition going around asking someone to finally write a few lines about heartbreak or drinking poison because the other person did it (which is dumb in my opinion)

And before someone built the first spreadsheet, nobody was probably sitting around thinking, if only there were a grid of cells that automatically calculated things for me. Well, perhaps someone was. I mean, accountants have always been a weird breed.

Look, the thing is, a lot of what we now consider essential started as an idea that probably seemed unnecessary to most people at the time. Nobody was asking for it because nobody knew what life would be like with it.

We’re quite good at asking for improvements to things we already do understand. We know what annoys us about our current music app or social media app, which part of our workflow takes too long, or why the software we use makes a simple task unnecessarily complicated. Ask us things that could be faster or cheaper, and we’ll have a list of suggestions ready.

But asking someone whether they’d like to read a poem that puts a feeling they’ve never been able to explain into words is a different thing altogether. The same goes for a product that changes how they approach a problem they thought they’d have to live with.

How are you supposed to ask for something when you don’t know it exists?

Sometimes, the person making something has to get a little ahead of the people they’re making it for. Not because they know better than everyone else, necessarily, but they’ve noticed something worth exploring and want to see where it leads.

Of course, this is also basically how we end up with 17 productivity apps that all promise to help us get our lives together. You see, having an idea doesn’t automatically make you a visionary. Sometimes it’s just an idea, and sometimes the world would genuinely be better off without it.

Still, somebody has to be willing to try.

The poet and the product builder have the same problem

I’ve always found it funny how much work goes into making something that, if you’ve done it right, barely feels like work to the person experiencing it.

Take a poet trying to describe the sadness of leaving a place you no longer belong to (Basically the first half of Meiyazhagan) . They might spend weeks on it, changing words that seem perfectly fine, deleting entire stanzas, then bringing them back because the original was somehow better. Eventually, they publish twelve lines that someone reads in thirty seconds.

Now imagine someone building a product. They’ve noticed that people keep losing track of what they’ve learned every time they switch between apps. (See what I did there… you know… cos lmnts.app…anyway...) They spend weeks thinking through the problem, testing different approaches, removing features, and rewriting things that looked perfectly good yesterday. After all that, someone uses the product and gets what they need in half a minute.

The creator has spent weeks thinking about an experience the other person barely has to think about at all.

A good poem doesn’t announce how difficult it was to write ( a bad poet might). It just makes you feel something, and you might never stop to wonder how many terrible drafts the poet had to get through to make those twelve lines work.

A well-made product is much the same. You click a button, find what you’re looking for, and carry on with your day. You don’t think about the decisions behind it, or the argument the team probably had about whether that button should even exist.

Most of the work is invisible to the person on the other side. Maybe that’s part of the fun of making things, even if nobody is going to send you a thank-you note for getting the spacing or the font right.

Most people won’t care. That’s part of the deal.

Now, here’s the less romantic side of making things.

You spend months writing a collection of poems, finally publish it, and discover that eleven people have read it. Three are friends, two are relatives, and one is your mother, who has read it twice because she thinks the first five readers might include you.

You spend months building a product, launch it to the world, and eleven people sign up. Three are friends, two are relatives, and your mother somehow manages to count as two users.

Ya laugh it up…cos It’s funny until you’re the person who made the thing. When you’ve spent that much time on something, it’s hard not to feel that other people should care about it too. I mean, why wouldn’t they, right? You’ve thought about it for months. You worked on it with such care, and you have spent entire nights tweaking the UX or Fixing the bugs…others have to care…right?

Then you put it out into the world, and the world doesn’t care as expected.

The internet makes this particularly strange. We can now announce our creations to billions of people, and most of them will never know we exist. They have their own problems, their own interests, and approximately forty-seven other tabs open. Your creation is competing with all of that, and nobody is obliged to give it a chance just because you worked hard on it.

Sometimes the thing you’ve made isn’t very good. Sometimes it’s good, but nobody knows it exists. And sometimes people understand it perfectly well and simply don’t need it. It’s tempting to assume the world just isn’t ready (trust me!), but that’s a rather convenient explanation to reach for every time something doesn’t work.

You have to leave room for the possibility that you got it completely wrong.

That’s probably one of the harder parts of creating anything. You need enough conviction to spend months working on something that might go nowhere, while accepting that your conviction doesn’t count as evidence that other people will want it.

The poet can write another poem tomorrow. The product builder can try another version. Neither has any guarantee that the next attempt will work out.

And honestly, what else can you do? You make the thing, put it out there, and see what happens.

But then someone finds it

Every now and then, someone comes across something you’ve made and gets it in a way you couldn’t have planned for.

They read a poem and think, Wait, other people feel this way too? Anyone who’s ever found themselves in a book or a song knows what I mean. It’s a strange thing to feel understood by someone who has no idea you exist.

Or someone starts using a product and realises they’ve been putting up with an annoying little problem( like how switching learning apps make you start things all over again) for years when they didn’t have to. Maybe it doesn’t transform their life. It just saves them half an hour every week, or means they don’t have to repeat the same frustrating process for the hundredth time. Small things, really, until you imagine going back to how you did them before.

The person who made the thing might never know what it ended up meaning to someone else.

The poet was probably thinking about a particular feeling when they wrote those lines. The product builder was trying to fix a particular problem. Neither could have known exactly who would eventually find their work useful, or what that person would take from it.

And then, somewhere along the way, the thing takes on a life of its own. Someone reads the poem years later and connects it to an entirely different experience. Someone uses the product in a way the original builder hadn’t anticipated.

I find that quite lovely, actually. You can spend all that time making something for reasons that are entirely your own, and then someone else comes along and finds their own reason to care about it.

You don’t get to control that part. You just hope the thing finds its way to someone who appreciates it.

Not everything needs to justify its existence

Of course, poetry and products aren’t quite the same thing.

A poem can exist simply because someone wanted to write it. It doesn’t need a business model or a retention strategy. Three people could read it, and it might still have been worth writing. There’s something nice about being able to make something without having to prove that it’s useful.

A product has a slightly harder time with that argument. If nobody finds it useful, you can’t keep insisting that people just don’t understand your vision. (Again…trust me, I have tried.) Eventually, you have to accept that people have other things to do and your app isn’t one of them.

Fair enough. Products do have to earn their place in people’s lives.

But I wonder if we’ve become a little too accustomed to asking what everything is for before we let ourselves get excited about making it. What’s the market? Who’s the audience? How much value does it create? When are you going to stop with this product bullshit and get married?

These are reasonable questions, particularly if you’re trying to build a business and would like to pay your bills.

Still, not every interesting idea starts with a neatly defined problem and a spreadsheet proving it’s worth pursuing. Sometimes you notice something that bothers you, or you come across a possibility you can’t stop thinking about. You don’t know whether it’ll work, and you certainly don’t know whether anyone else will care. You just want to see if there’s something there.

I think there’s room for that kind of curiosity, even when you’re building something practical. You don’t have to know exactly what you’re going to end up with before you begin. Sometimes you only figure out what the thing is supposed to be by trying to make it.

Of course, at some point, you have to stop listening to your own enthusiasm and pay attention to the people you’re making it for. Being attached to an idea doesn’t make it a good one. If you’re building a product, you have to be willing to change things when reality doesn’t quite match your expectations.

But that doesn’t mean the original impulse was pointless. You noticed something, followed it, and found out what was there. Even if the result wasn’t what you expected, at least you know a little more than you did before.

I don’t think every poet sits down to change the world, and I doubt every product builder wakes up thinking about their contribution to humanity. Sometimes you just have an idea that won’t leave you alone, and after a while, doing nothing about it becomes more annoying than trying.

So you write the poem. You build the product. You spend an unreasonable amount of time getting something right that most people will probably never notice, then put it out into the world without knowing whether anyone will care.

Usually, life goes on. A few people might like what you’ve made. Most won’t come across it at all, and that’s probably how it should be. We can’t all spend our lives paying attention to everything everyone else creates.

But occasionally, someone finds the thing at exactly the right time. They read a few lines that put a feeling into words they’ve been searching for, or they use a product that makes an everyday frustration disappear. Something that began as an idea in a stranger’s head ends up making a small but genuine difference to someone else’s life.

The creator might never find out. They could be off writing another poem or building something else entirely, having no idea that the thing they made has become part of somebody else’s story.

And maybe that’s enough. You don’t have to know exactly who will find your work, or what they’ll make of it. You just have to care enough to make it in the first place, and be willing to let it go once it’s out there.

Nobody asked for another poem. Nobody asked for another product. But the world would be a very dull place if we only ever made things people already knew they wanted.

Sometimes, someone makes something anyway, and years later, someone else is very glad they did.

I think that’s a good enough reason to keep making things.

And my Claude Code session just reset. So, yeah. Back to building.

I got diagnosed with moderate ADHD a few months ago, and my first reaction was basically, “Ohh, so that’s why I do that.”

The more I thought about it, the more obvious it got. A bunch of things I’d always written off as personality quirks suddenly started looking like patterns.

I’ve always loved starting things. I’d take on a project, pick up a hobby, get interested in some random technology and convince myself I’d stick with it this time. I’d get obsessed for a week, learn far too much, and somehow end up reading about the life of the person who created the framework instead of actually using the framework.

I forgot small things. I put off things I genuinely cared about. I kept telling myself I’d get everything done at the last minute, and sometimes I did. Sometimes I absolutely didn’t.

When I was younger, nobody really cared. People said I was “scattered” or “curious,” and some people even found it kind of fun. A “quirk.” I did too. It was just part of how I was.

Then I co-founded a company, and it stopped being charming pretty quickly.

At a job, there’s usually some structure around you. Someone notices if you vanish into a rabbit hole for three days. A deadline doesn’t care that you suddenly got interested in something else. A team expects you to finish what you said you’d finish.

In a startup, most of that disappears.

You and your co-founder decide what matters, what to build next, whether the 1 AM idea is worth chasing and what gets ignored for now. You’re also still supposed to send the follow-up email, check the numbers, deal with the boring operational stuff and keep track of all the little things that keep a company moving.

You’re sharing the steering wheel with someone while also being the person who forgot where you parked the car.

I eventually had to step down from the company I co-founded, which is still a weird sentence to type.

For a long time, I filed the whole thing under “I wasn’t disciplined enough.”

Discipline does matter. But I was also trying to build a company with a brain that works differently, in a world that has gotten remarkably good at feeding exactly what that brain can’t resist.

the circus

Everyone has a bit of a monkey brain. Mine is a monkey that’s had coffee and Red Bull and somehow found a magic lamp.

Everything is one tap away now. You can order something and have it at your door before you’ve forgotten why you wanted it. You can buy a domain in two minutes and put a landing page on it.

I’ve owned an embarrassing number of domains for ideas that lasted less than a weekend.

Coding is probably the worst one for me, because I love it.

There used to be a lot of friction between having an idea and doing something about it. You had to set up the project, pick the stack, read the docs, configure everything and inevitably lose an hour trying to figure out why some stupid thing wasn’t working.

It was annoying as hell, but looking back, it was also doing me a favour. It killed most of my bad ideas before they became projects. They’d usually die somewhere between “this would be cool” and “why is the database not connecting?”

Now I can describe an idea to an AI and have a working prototype while I’m wandering around the internet doing something completely unrelated. I use this constantly, and I’m definitely not giving it up.

The side effect is that almost every random thought can now become a real project.

Every new project gives me a feeling I really like. I’m building something. Maybe it’ll turn into something interesting.

The beginning is great because everything is still possible. There’s stuff to learn, directions to explore and no boring history attached to the project yet.

Then the exciting part ends. The project turns into maintenance. The interesting technical problem turns into an edge case. The prototype has to become a product. Someone has to write the documentation, answer the email, fix the small bug and the UX, and do the thing you promised you’d do.

My brain has never found that part nearly as exciting.

The monkey wants the start. It couldn’t care less about version 1.7.

the problem with a co-founder

This is the part I find hardest to write.

When it’s just you, your distraction mostly costs you. You start a hobby and abandon it? Fine. You start a side project and never finish it? Nobody really cares.

With a co-founder, it costs them too.A new idea would show up, I’d get excited, and I can be annoyingly convincing when I’m excited.

Nobody above us was going to say, “Interesting, but not now,” so it came down to the two of us deciding whether something was worth pursuing.

And the monkey is very good at making a shiny thing sound like a strategy.

Sometimes the idea really was good, which is what made this so hard. Some of them were worth exploring, and a few became genuinely useful.

Other times, something was just more interesting than what we were supposed to be doing. From inside my head, the two felt almost identical.

“Maybe we should pivot.”

“Maybe this is a better opportunity.”

“Maybe this is the thing we’ve been missing.”

At 1 AM, those sound a lot like vision.

Meanwhile, the boring things don’t feel urgent until they suddenly are. The follow-up I didn’t send. The UX I didn’t fix. The number I didn’t check.

None of them seemed important on their own, and somehow they all came back bigger. And when you’re building something with another person, whatever you drop doesn’t disappear. Someone else has to pick it up.

That’s what bothers me most when I look back. Some of the consequences of my distraction weren’t mine alone.

adding friction back

I don’t have a perfect ADHD system.

I’ve known about this for only a few months, so I’m still figuring out what actually works and what’s just another system I’ll get excited about for two weeks.

One idea keeps coming back, though.

If the world has removed almost all the friction between an impulse and an action, I need to put some of it back.

If I want to buy something, it goes on a list and I wait a few days. Most things don’t survive the wait.

If I get a business idea while I’m working on something else, I write it down instead of opening a code editor. If it still looks brilliant tomorrow, I can look at it again.

I’ve also deleted a lot of the stuff that makes it too easy for me to follow every random thought. I turned off notifications and made it slightly more annoying to reach anything I know will happily eat an hour.

None of this is particularly clever or revolutionary

What changed is that advice I’d ignored for years finally started making sense once I stopped treating it as a test of willpower and started looking at my environment.

It works in the other direction too.

For the stuff that matters, I want almost zero friction.

I want the first step to be stupidly easy. If I need to write something, the document should already be open. If I need to send an email, the draft should already be started. If something feels too big to begin, I shrink the first step until my brain has no good argument against it.

I don’t need to love boring things.

Starting just has to be easier than avoiding.

I’m also getting professional help, which I think is worth saying plainly. I spent years believing this was a willpower problem. It wasn’t.

I wish I’d figured that out sooner, although it’s hard to be angry at yourself for missing something you didn’t know was there.

the monkey stays

I don’t think I even want to “fix” it completely.

The same brain that struggles to sit on one task can also get completely obsessed with something and go way deeper than anyone asked.

It makes weird connections between unrelated things. It gets excited about learning something new. It can spend an unreasonable amount of time trying to understand something just because it suddenly became interesting.

I actually like that part.

A lot of the projects and hobbies I abandoned weren’t wasted either. Some taught me something useful. Some planted an interest that came back later. A few just left me with completely useless knowledge that will eventually come up in a conversation.

I want to keep that curiosity.

I just want to get better at noticing when I’m being pulled toward something and asking myself whether I actually want to follow it.

The circus is staying, by the way.

It’s bigger now, cheaper, and open 24/7.

There will always be another project, another hobby, another rabbit hole and another domain name that somehow feels like a very good idea.

The monkey can stay too.

I just don’t want to buy it a ticket to every show.

I have this strange theory… we are heading towards a world without poets… not poems, mind you… poets. Thanks to AI, we could still have poems everywhere. In fact, I think we might just have more poems than ever. But because of the way we teach things, I think POETS are facing an extinction-level crisis.

Now, this blog isn’t really about poetry. It’s just a convenient way to get the point across.

Today, you can ask AI to write a poem about anything and get something decent in a few seconds. You might have noticed this when that distant aunt or uncle suddenly developed a talent for creative writing and started posting all those weird WhatsApp statuses. You can give GPT a vague idea and watch it turn that idea into something that sounds like it took someone sometime to write.

I use these tools too (Not for writing poems, FYI), and I think they are amazing. And whenever I use it, I feel strange seeing how fast one can get from having an idea to having the finished thing.

I think that gap between the idea and the finished thing matters and I was thinking about this gap after watching a recently popular lecture by a professor from IISc Bangalore, who starts the video by asking his students why they still come to class.

I mean, his point is pretty straightforward. The lectures are recorded, the material is available online, and now every student has access to AI tools that can sit with them and explain the same topic in five different ways, and it never get annoyed or cut their grades when they still don’t understand it…so why come to class?

You see, for a long time, one of the main jobs of a teacher was to explain stuff. The teacher knew something you didn’t, so they gave you the explanation from a properly curated textbook. Now, that made sense when precise knowledge was difficult to find.

Well, that world has changed, and that particular model of teaching is now useless. Think about it, a student can now have a remarkably patient tutor sitting inside their laptop, ready whenever they are. And I’m pretty sure it is already better than most teachers at explaining things, and I don’t think the correct response here is for teachers to get pissed or somehow try and compete with it.

The professor in the video makes a distinction that I found very interesting and relevant. He talks about the goal of education being competency rather than simply remembering particular materials or facts from a course. He says he wants his students to reach a point where, years later, they can pick up a paper that they have never seen before and understand it themselves. For him, the course he is teaching is simply a way to develop that ability.

That is a very interesting perspective, and it got me thinking about what we actually mean when we say someone has learned something. I think all of us would agree to the fact that we tend to measure learning via the thing a student produces at the end. They write the exam, or they do their homework or submit the project and get a grade, and that grade somehow represents how well the student has learned something.

For a long time, that was a reasonable proxy because producing the answer usually required the student to spend time on it and develop the ability to get to that answer. Well, AI has thrown a huge brick through that system.

A student can now get a perfectly reasonable essay without necessarily becoming a better writer or a sharp thinker. They can get a working piece of code without understanding the decisions that produced it. And they can even ask for a poem about heartbreak without ever having to sit with a blank page (or a broken heart) and figure out what they actually want to say.

All they need now is the right prompt and the output can be there even when the experience that was supposed to produce the learning never happened. And that is much more alarming than whether students are going to use AI to cheat. Because I think we have underestimated what the PROCESS of making something does to the person making it.

Think about writing poems. You might read a beautiful poem and think, “That’s incredible.” But your relationship with that poem changes after you try to write a poem yourself. You start noticing why a particular word or phrase was chosen. You understand how difficult it is to come up with a metaphor or how hard it is to write something meaningful and make the words rhyme at the same time.

I mean, you don’t have to become a great poet to appreciate poetry. You just need to have some relationship with the process of writing.

The same thing happens with art. Someone who has spent years trying to draw will look at a painting differently from someone who has never picked up a pencil. They might notice a decision about color or composition or light that another person completely misses.

I think that is what learning does at its best. It changes the way you see things.

Now, the problem is that AI has created this strange situation where one can now skip the effort and jump straight to the result.

I do think there can be situations where it is useful to get quick results, but education is different. When you are learning things, the process itself is often the point.

The problem is that our current education system doesn’t always make that very obvious. We give students assignments where the thing being assessed is the finished product, and then we are surprised when students use the most easiest way available to produce that product.

Think about it: if the assignment is supposed to teach them how to write but the only thing being checked is how the essay looks at the end or how many pages it has, why would they spend five hours struggling with the writing? And this is not necessarily a problem with the student.

The professor in the lecture gets to this exact issue when he talks about open-ended assignments. With a large class and everyone having access to AI, the final answer tells the teacher almost nothing about what actually happened during the learning process.

This is where I think teaching has to change.

The answer probably isn’t to pretend AI doesn’t exist. The professor himself suggests using AI as part of the learning process and then using class time for interaction and discussion.

That makes much more sense to me.

Let the machine explain the basics. Let students ask it stupid questions without feeling embarrassed. Let them use it to explore things before they come to class. Then make the classroom the place where they have to actually engage with the subject. They should have to defend an idea they arrived at. They should have to notice when an answer doesn’t make sense. They should have to wrestle with a problem that doesn’t immediately give them a clean path to the solution. The teacher’s role becomes creating those situations and knowing when to push the student a little further.

That is a very different job from standing in front of a classroom and just reading from the textbook, hoping that the students learn something.

The professor makes another comparison in the lecture that I liked. He suggests that universities(and schools) might eventually become more like gyms. You can exercise at home, but there is something about being in an environment designed for the activity that changes your behaviour.

I think learning works in a similar way.

You can get information anywhere now. What is much harder to get is an environment that makes you stay with something long enough for it to change you. That might actually become the most valuable thing a teacher provides.

The teacher doesn’t have to be the person who knows all the answers anymore. AI is probably going to be better at that anyway. The teacher has to help create the person who knows what to do with the answers.

Given how easy it is to generate stuff now, i think we might see a lot of beautiful looking things out there, but we might end up with fewer people who has developed a deep understanding of why certain things work or why they make us feel something.

And that is the part that worries me.

Imagine growing up in a world where almost anything you want can be generated with a prompt. You would still have all these beautiful things around you……But would you appreciate it in the same way?

I think there is something about attempting to create things by yourself that gives you a deeper relationship with creativity. You start to understand how hard it is to get to the finished product and without that experience, the effort can become invisible. And if the effort becomes invisible, It’s super easy to take things for granted.

That is why I keep coming back to the poet.

A Gibran or a Wordsworth isn’t interesting simply because they could produce a poem. AI can already do that, and it will get much better at it. What makes them interesting is everything that had to happen before the poem existed: years of experience, their outlook of life, their writing, failing, paying attention to language, and slowly developing a sense for why one line feels right and another doesn’t. The poem is the thing we see at the end, but the real interesting part is the person and their mind, who had to become capable of writing it.

If teachers don’t change how they teach, I worry that students will simply get very good at delegating those experiences to ai. They will learn that whenever something feels difficult, there is probably a GPT that can do it for them. Over time, the education system can become very efficient at producing finished work while becoming less effective at developing the people who are supposed to make sense of that work.

And eventually, the prompt becomes the effort. And when that happens, we might have kids who see a frame of a Ghibli film and say something like, “whats so special about that, i can do the same thing on chatgpt”.

See…I don’t think the answer here is to keep AI out of education. I think it should be the opposite. I think, we should let AI handle the parts of learning that machines are now exceptionally good at, and then redesign the parts that were never really about just getting the answer.

Teachers need to create experiences that make students think, and eventually see something they couldn’t see before. The assignment has to matter because of what happens to the student while doing it and not simply because something gets submitted at the end.

Otherwise, we may accidentally teach an entire generation that learning is just the process of getting to the answer as quickly as possible.

And if that happens, I don’t think poetry will disappear.

We’ll probably have more of it than ever.

I’m just not sure we’ll have as many poets.

On borrowed thoughts, half-finished ideas, and how much of ‘me’ AI could actually recreate

So I have this yearly ritual. At some point during the year, I go back and watch a few films that have stuck with me over the years. And when I say watch, I mean actually watch the whole thing. Not 1.5x through it.

One of my favourite films in this little ritual is Anand Gandhi’s Ship of Theseus, a 2012 film that is available for free on YouTube.

Every time I watch it, I end up thinking about something different. The film is built around the old Ship of Theseus thought experiment: if you keep replacing the parts of something, at what point does it stop being the same thing?

The basic version is pretty simple. Imagine a ship whose parts start breaking over the years. You start replacing the parts with new ones. Eventually, every original part is gone. The ship is still sailing and still looks like the same ship, but none of the original material is left.

So, is it still the same ship? (Listen… if this bores you, I suggest you skip the rest…)

I have always found that question interesting, but this time around I started thinking about something completely different.

AI.

More specifically, I started wondering how much of what I think is actually my own thinking anymore.

I use Claude and GPT and stuff for almost everything now. I mean, not literally everything, but enough that I sometimes start opening a chat before I have even finished forming the thought in my head.

Initially, it was pretty basic. “Fix this email.” “Explain this error.” “Make this paragraph less AI.” Pretty harmless stuff.

Then it became brainstorming….Research…. Asking it to challenge an idea……Asking it to tell me why my idea was probably wrong… Ghibli images and the 80s trend. Eventually, I was having two-hour conversations about things I hadn’t even properly figured out yet.

Somewhere along the way, the model started becoming part of how I actually think.

That is a slightly weird realisation.

You have some vague thought in your head, you type it out, and the model gives you a much clearer version of it. Maybe it adds something you hadn’t considered or connects two things you hadn’t connected before. You read the response and think, “Yes. Exactly. That’s what I meant.”

Except it wasn’t. The model actually helped shape that thought.

So where does that thought belong? (Second warning…you can skip the rest if you like)

We have always borrowed ideas from other people. Most of what I believe came from somewhere. A book, a conversation, something a friend said, something I learned at work, some random experience that stuck with me for whatever reason. I mean, I don’t think any of us are particularly original thinkers. Our “thoughts and views” are usually collections of things we’ve picked up from other people and the world around us.

But AI makes the process much more direct.

I can sit with a model for three hours and essentially think out loud. I can give it a half-baked idea, ask it to challenge me, argue with it, tell it that its answer is wrong, ask it to try again, and eventually arrive somewhere I probably wouldn’t have reached on my own.

And then I started wondering about something even stranger.

What if the model had enough of me?

Let’s say I gave Claude or GPT years of my WhatsApp conversations. The memes, the random messages, the arguments, the way I talk when I’m annoyed… all of it. Basically, all the little pieces that make up the person my friends think they know.

And then I give the model access to WhatsApp and let it have a conversation with one of my close friends.

Would they know?

My immediate reaction is obviously yes. Of course they would know.

But then I think about it a little more, and I’m not so sure.

My friends know things about me that aren’t written anywhere. They know the context behind a sentence. They know the things I don’t bother explaining because we’ve known each other long enough that I don’t need to.

But a surprisingly large amount of who we are online is made up of patterns. The words we use, the things we talk about, the arguments we repeat, what annoys us, what excites us, the weird things we obsess over, even the way we respond when someone disagrees with us.

If a model had enough of those interactions, maybe it could get pretty close.

No, I’m not talking about an AI actually being me. That’s a completely different question. But can it produce something that my friends would recognise as me?

And that is where the Ship of Theseus started making a lot more sense to me.

There is another version of the thought experiment where the old parts of the ship are collected and eventually used to build another ship. So now you have two ships. One has the continuity of the original ship, while the other has all of its original parts.

Which one is the real Ship of Theseus?

I don’t know.

But it made me think about all the pieces of ourselves we leave behind online: our messages, writing, code, emails, photos, opinions, conversations and thousands of tiny interactions that don’t seem particularly important individually but probably reveal quite a lot about us when put together.

What happens when you give all of those pieces to a model?

Can it build something that interacts like you?

I don’t mean consciousness and all that. I’m not suggesting there is going to be another Sethu sitting inside a server somewhere, waiting to take over my WhatsApp (I hope)

I mean…could it interact like me? Enough to confuse a few of my friends at least….

If you gave it enough material, I think the answer might eventually be yes.

And that is a pretty strange thought.

We usually talk about AI learning things about us as if it is learning facts. What I find much more interesting is the possibility of it learning our patterns. And if it gets enough of those patterns, how close can it actually get to the person?

I’ve been thinking about this especially when I use these models to form an opinion.

If I read a book and it changes my mind, I still consider the new opinion mine.

So why should AI be any different?

I don’t really know.

I think what gets me a bit more is how quickly I can go from having a vague feeling about something to having a reasonably well-formed opinion about it. I can throw the thought into a model, go back and forth for an hour, and come out with something that sounds like I’ve spent weeks thinking about it.

Sometimes I have……Sometimes I definitely haven’t.

And because the final version sounds much better than the original mess in my head, it’s pretty easy to forget that the mess was where I actually started.

Maybe that is something worth being careful about.

…No.. i am not talking about abandoning the use of AI all of a sudden and going back to typing 500 lines of code with your hands or spending weeks on research reports (god no!). I was simply wondering what happens when you get so used to having something available to bounce every thought off that sitting with an unfinished thought starts feeling unnecessarily difficult.

There are times when I probably should just sit there and think.

Instead, Now I simply open a chat.

We have always thought with other people anyway. We absorb things from the world, argue with our friends, read something that changes our mind, steal an idea from someone else, and eventually forget where it came from. That is just how thinking works.

The difference is that now one of the things we are thinking with isn’t a person.

And unlike most people, it is available at 3 am and has absolutely no problem listening to me spend forty minutes trying to explain a thought that probably wasn’t worth forty minutes in the first place.

Which brings me back to the movie.

The movie is about identity, change, and how much of something can change before we stop calling it the same thing.

And AI makes that question a little more interesting.

Because we are not just changing ourselves anymore. We are increasingly thinking alongside something else. We give it our ideas, it gives us ideas back, we accept some, reject others, change our minds, try again, and eventually something new comes out of the conversation.

So if we give a model enough of our thoughts, our writing, our conversations and all the weird little patterns that make us us, and it can build a ship that our friends recognise as us, I mean, at that point I guess we have to wonder what exactly makes my ship mine in the first place.

Introducing LSL, an open format for representing the current knowledge state of a learner

Learning is cumulative. Every concept you understand builds on the ones before it, gradually forming a picture of what you know. Today, the surface of learning is expanding super fast. We learn things from across courses, documentation, YouTube videos, coding platforms, AI assistants, and conversations with language models. Every one of these interactions contributes something to our understanding of a subject.

Many of these learning systems keep an orderly record of what a learner did. They use standards like xAPI or Caliper to log learner activities, and these logs do an excellent job of answering questions like: What did the learner do or Which quiz did they complete? Which video did they watch? Which lesson did they finish?

What these records can’t answer are more direct questions, like: what does the learner know right now?

Every one of these learning platform builds its own inferred picture of the learner, and the problem is that this context of the users understanding rarely leaves the platform that created it. So every new tool asks you to begin again, reconstructing context that already exists somewhere else.

Over the past few weeks we have been thinking about this problem. As AI becomes a huge part of how we learn, a portable learner context starts to feel less almost like a necessity. I mean, the question is obvious, What if the understanding you have built around a subject wasn’t tied to the platform that you used to learn it?

These questions led us (Myself and Anantharaman) to working on something we call, LSL (Learner State Layer). It is an open format for representing a portable user knowledge state. Instead of recording a history of learning events or activities , LSL captures the learner’s current knowledge state. It is basically a snapshot of what a learner knows, where they are struggling, the misconceptions they are carrying, and how confident the system is in each of those observations.

This blog is an introduction to the idea, why we built it, and where we hope it goes.

Why we built it

Every tool you learn with is quietly building a detailed picture of what you know, but almost none of them will let you take that picture anywhere. A coding site helps you develop a real sense of where you are strong. A language app notices words you keep confusing. The AI assistant you study with slowly works out how you think. All of that understanding exists, and all of it stays sealed inside whichever platform happened to observe it, written in a private format nobody else can read.

What the learner gets in return is a kind of amnesia among these platforms. You spend a year getting genuinely good at algorithms on one platform, then you sign up for another the following week, and get treated like a complete beginner, onboarding test and all. You study with an AI for months and still watch every fresh conversation open at zero.

The knowledge you have built is real, but the ability to carry it with you across platofrms is the piece that has been missing.

That is what LSL is meant to supply.

LSL is short for Learner State Layer, and the idea behind it is deliberately modest. It is a single file, owned by you, that describes what you currently know, and it can move with you from one tool/platform to the next. We call the broader concept a portable user knowledge state.

What is actually in the file

LSL isn’t a transcript or an activity log. It doesn’t record everything you have watched, or completed. Instead it sits on top of those signals and represents a snapshot of what you know right now, so that any learning system can use that snapshot before deciding what to teach, explain, or assess next.
A small slice of one an LSL looks like this:

lsl: "1.0"
learner: "ben"
updated: 2026-07-08
subject: "Data Structures & Algorithms"

concepts:
  - name: "Memoisation"
    mastery: 0.68            # a rough sense of current ability, from 0 to 1
    confidence: med          # how much to trust that number
    last_seen: 2026-07-08    # knowledge fades, so freshness matters
    can:
      - "Writes a correct top-down memoised solution for standard problems"
    cannot_yet:
      - "Analyses the space trade-offs of memoised versus tabulated solutions"
    misconceptions:
      - "Believes memoisation always reduces time complexity to O(n)"

As you can see, some of those claims describe demonstrated abilities, such as “writes a correct memoised solution for standard problems.” Others capture areas that are still developing, like “still exploring the space trade-offs between solutions”. The goal here is to describe the learner’s current state in plain language.

Another interesting aspect of a learner that LSL captures is there misconceptions. LSL is also designed to record the incorrect mental models a learner currently holds. A claim such as “believes memoisation always reduces time complexity to O(n)” might look like a minor detail, but it could change how an intelligent tutor should respond.

A tutor working only from quiz scores knows that a learner struggled with a topic and will probably explain it again. A system that understands the learner’s misconception knows why they struggled, and can challenge that specific false belief before it becomes the foundation for further confusion.

You can find the complete spec in our official github repository.

Why this is possible now

The idea of moving learner data between platforms isn’t new, and plenty of standards have tried it before. Most of them struggled for the same reason: They where all designed for rigid programs/applications and every system had to agree on exactly what every concept meant. If one platform wanted to record “recursion” and another wanted to read it, both had to use the same predefined identifier or vocabulary. Creating those shared vocabularies took years, and keeping them current was harder still.

LSL assumes that the reader is no longer a traditional program that can only match predefined codes. The reader is an AI, a system that can actually reason about language and that change relaxes almost everything.

Instead of forcing every platform onto identical concept IDs or capability names, LSL lets each one describe what a learner knows, doesn’t yet know, or misunderstands in plain language, and trusts a modern language model to interpret those descriptions much as a human tutor would. One platform might call a concept “linked lists” while another calls it “list nodes.” An AI reader recognises that they are talking about the same idea, without either system having to agree on a shared vocabulary in advance. By leaning on reasoning rather than rigid identifiers, LSL stays simple, and sidesteps most of the standardisation work that made earlier learner-data formats so painful to build, maintain, and adopt.

What it refuses to do

When people first hear about LSL, one question comes up almost immediately: can any app just claim that I know something? It is a good question, and the format is designed to prevent precisely that.

Simply watching a video earns you nothing. If all you did was sit through a lesson, the file cannot suddenly claim you know how to do anything. Capability claims require evidence, which means you have to actually demonstrate the skill. That is a deliberate move away from the old assumption that finishing a course automatically proves competence.

Alongside each concept, LSL also records a confidence level and the date it was last seen, both attached to its estimate of how well the learner knows that concept. That is because knowledge changes over time. Someone who demonstrated a skill yesterday sits in a different position from someone who last demonstrated it six months ago.

We also make a very important distinction. A basic LSL file is not a credential, and it isn’t meant to stand in for a certificate or a degree. Think of it more as a structured summary of what you currently know. That is plenty for a learnng system, which can ask follow-up questions and verify any claim before teaching you, and nowhere near enough for high-stakes decisions like hiring or admissions.

For those situations, LSL has a heavier optional layer of its own, that adds cryptographic signatures and a record of who issued each claim, so a document can be trusted even as it moves between organisations. And when what you actually need is a formal credential, a certificate or a degree, that job belongs to standards built specifically for it, such as Open Badges, which can happily point back to an LSL claim without LSL ever pretending to be the credential itself. The basic format stays intentionally lightweight, and the specification discourages stretching it to cover purposes it was never designed for.

Why any of this matters

The real reason we built LSL is simple: better personalisation.

Today, “personalised learning” usually means a platform rearranges its own content based on what you have done inside that platform. The moment you leave, everything it learned about you stays behind.

LSL is built around a different idea. Imagine opening a new AI tutor for the first time, and it already knows you are comfortable with recursion, still learning tabulation, and carrying one specific misconception worth fixing before you move on. There is no long onboarding, no diagnostic quiz, and no guessing. The tutor can begin in the right place because it already understands your current knowledge.

Since the LSL file belongs to you rather than to any one platform, that understanding travels with you. Your coding tutor, your language-learning app, your online course, and your AI assistant can all build on the same picture of what you know, and that picture grows more accurate as you learn instead of being recreated from scratch every time you switch tools.

You decide which apps can read your LSL file, and you can read it yourself, because every entry is written as a clear, human-readable statement rather than a hidden score quietly shaping what a system shows you next. That readability matters most for the sensitive parts, like your gaps and misconceptions, which is exactly why the file is yours to share on your terms.

Where things stand

We also want to be clear about where LSL stands today. This is an early public draft, and there’s still a lot to validate. Some parts of the specification, especially things like automatically identifying misconceptions, are genuinely difficult problems. Many early implementations will probably do something much simpler: provide a clear picture of what a learner currently knows. That’s a useful place to start.

What keeps us excited is the core idea itself. Once you hear it, it feels obvious. Your learning shouldn’t be locked inside the platforms you use. It should belong to you, and the tools you learn with should be able to understand your current knowledge from the start.

The specification is open, and we’re publishing it because we want feedback. We’d much rather hear where it breaks than be told it’s a good idea. If you build learning tools, teach, study, we’d love for you to read the spec and tell us what we’ve missed.

This is just the first version. Where LSL goes from here depends on the people who build with it, challenge it, and help improve it.

You can learn more about LSL from our official website: learnerstate.org

Ok so I was today years old when I found out that the whole “I’m a visual learner” thing I’d been saying since like the eighth grade is… not really a thing.

And look, I know. A lot of you are going to tell me that you definitely are a visual learner and that pictures genuinely help you learn better, and honestly, yes, fine, I want to believe you. But stay with me for a second because I just found out that you are basically lying to yourself to look cool.

So somewhere in school, someone told all of us that people have fixed learning styles. Visual, auditory, kinesthetic, etc . The most popular version of this, the VARK(Visual, Auditory, Reading, Kinesthetic) model, came out in 1987, and it took off and It spread because it feels true.

Some people love videos. Many love reading. These preferences are real, and I’m not arguing with that. The issue is what we decided those preferences meant.

Somewhere between “I prefer this format” and “I can only learn this way effectively,” something quietly went wrong. And the gap between those two statements is where forty years of education policy went to live.

Researchers went looking for actual evidence, took visual learners, taught half visually and half differently, see who retains more. The studies found no significant difference. A follow-up study watched university students through a whole semester and found that most people weren’t even studying in their own self-identified style. The ones who were didn’t perform any better.

96% of teachers still believe in learning styles. The belief is everywhere. But ya, there is literally no evidence that these “Styles” help.

I’d been using “I’m a visual learner” as an excuse for years without fully realising it. Looking back, it was an incredibly convenient explanation because it meant the problem was never the effort. It was always the “delivery method” or the “format.

The research basically looked at all of this and said: “Yeah… we’re gonna need you to stop doing that.”

Because what actually seems to matter was the kind of effort you’re forced to put in, especially the annoying kind.

Retrieving something from memory beats rereading it. Spacing practice over time beats cramming. Testing yourself beats passively reviewing notes. Again and again, the things that feel harder in the moment tend to be the things that stick.

Which is honestly rude and exhausting.

The really annoying part is that the methods that feel the most effective are often the least effective.

Have you ever watched a beautifully produced explanation video and felt like you’ve completely understood a topic?

Of course you have.

For the next ten minutes, you’re practically an expert.

Then three days later, someone asks you about it, and your brain returns a vague memory of the thumbnail and some background music.

It turns out that understanding something while you’re looking at it and remembering it later are two very different skills.

There’s also the reality that subjects have opinions about how they want to be learned, and your preferences don’t mean shit. Languages want you to speak. Maths wants you to solve problems. Programming wants you to write code and then discover that the bug was caused by something so stupid you’d rather not talk about it.

The subject doesn’t care whether you’re a visual learner, an auditory learner, or a kinesthetic learner. It has requirements.

And unfortunately, most of them involve doing the hard things.

so.

People are genuinely different as learners. Prior knowledge, anxiety, motivation, time, these things affect how learning lands and always have. The problem was never that people have different preferences. The problem was the conclusion we drew from it, that preferences are limits, that if something isn’t clicking, the format must be wrong for our brain type rather than the thing just being hard.

It gave a lot of us a comfortable exit from the discomfort of learning difficult things. Most of us took it without questioning it.

I definitely did.

but ya, now you don’t have any excuses. Just accept the fact that learning things is hard and try harder!!

The Recipe for Internet Fakery

Welcome, aspiring fraudsters and con artists! Congratulations on deciding to abandon your real identity and join the lucrative world of internet deception.

You’re in excellent company.

I mean, there are literally millions of bots, fake artists, and non-existent photographers out there. And even one very successful Swedish guy who somehow became 656 people at once.

This guide will teach you everything you need to know about becoming completely fake online while making crazy amounts of money and facing absolutely no consequences whatsoever.

Because let’s be honest, that’s kind of the world we’ve built at this point, and you’d be stupid not to take advantage of it.

So let’s talk about it.

Step 1: Understand the Business Model

The entire internet runs on one simple principle: being fake is more profitable than being real.

And I mean significantly more profitable. Like “why would anyone even think of being real anymore” levels of profitability.

Let’s start with YouTube. Specifically, the goldmine that was AI slop before YouTube (mostly) killed it.

Adavia Davis is a 22-year-old college dropout who runs five YouTube channels. He makes roughly $40,000 to $60,000 a month. That’s $700,000 a year.

He built this whole network of channels churning out AI-generated content. Minecraft compilations aimed at kids. Funny animal videos. Prank compilations. Celebrity gossip. Anime edits. etc. None of it is original.

The man spends two hours a day overseeing the whole operation, and operating costs go to about $6,500 a month for small teams managing different niches. The Profit margins are around 85–89%. That’s actually better than most actual tech companies with large number of “professional” 9-to-5ers.

YouTube loved it because their ads still ran. Advertisers got their impressions. Davis got paid. Everyone won except the people actually trying to make real content, but who cares about them.

Then YouTube caught on.

On July 15, 2025. They updated their monetisation policy. The new policy started targeting “inauthentic content”, meaning mass-produced, AI-generated videos with no real human value added.

Channels posting repetitive AI slop started getting demonetized.

One channel called “True Crime Case Files” with 83,000 subscribers got completely removed after posting 150+ AI-generated murder stories. Another network got caught spreading AI-generated fake news about celebrities using deepfake images and AI narration.

But Davis is still making money. Because he adapted. He added just enough human input to stay on the right side of YouTube’s new rules. Not much. Just enough.

So, the lesson here is that these platforms will always try to catch up. But if you’re smart, you can stay one step ahead.

Now, let’s look at the numbers and see what all this actually costs:

Traditional content creation means hours of research and scripting, followed by more hours of filming and editing. Then there are the equipment costs, Software subscriptions, and whatnot.

And after all that, if you are really, really hustling, you might be able to put out 3–5 medium-length (< 20 mins) videos a week.

Well, an AI slop factory can generate a video with AI in 10 minutes. Add robotic voiceover in 5. Upload in 2. That’s a total of ~20 mins of your life, and the best part is that it can pump out hundreds of videos per week.

Total costs may be $100/month in tools and maybe a few extra dollars for chips and refreshment while you wait for AI to generate your stuff.

Before July 2025, people were making millions doing this. After July 2025, people are still making millions doing this. They just had to get slightly more creative about it.

So just be really smart and creative about it, that’s the business plan!

Step 2: Exploit the Psychological Loophole

Now, you might be thinking: “But won’t people feel bad about being lied to?”

HAHAHA. No. Not really.

Psychologists have a term called “self-discrepancy theory,” meaning there’s a gap between who you actually are (boring, flawed, human) and who you wish you were (interesting, successful, worthy of love).

The internet was supposed to help us close that gap through genuine connection and self-improvement and YouTube videos and all that stuff.

Instead, we realised we had free will, and we could just pretend to be all that, and nobody could actually prove otherwise.

I mean, about 62% of social media users admit they present a version of themselves online that’s very different from reality. And that’s just the ones honest enough to actually say it out loud.

Look, I know you know this, but the internet crowd isn’t looking for authenticity. They’re looking for the version of you that makes them feel something. Inspired. Jealous. Entertained. Horny. Whatever……

Studies show that 41% of people who create fake personas do it because they’re lonely. They want connection, but they don’t feel like their real self actually deserves it. So they build someone better. And the internet makes the whole thing really easy.

There is no eye contact needed, no weird body language and no awkward pauses where you don’t know what to say. You can edit out all the bloppers and the fumbles and delete the bad takes. You can curate your entire existence.

Psychologists call it the “online disinhibition effect” where people feel free to act in ways they wouldn’t offline. You can be braver, funnier, smarter, or just a completely different species of mammal if that’s what you’re into (like a dog!).

Pro tip: Every time you go into your fake persona, the gap between your real self and fake self gets wider, which creates more anxiety, which makes you double down on the fakeness even harder. It’s a beautiful self-sustaining cycle! You’ll never have to be your authentic self again! Yay!!

Step 3: Master the Revenue Model

Okay, enough with the philosophy and psychology and stuff. Let’s talk money.

Because sure, some people fake it for self-esteem or whatever. But as my man Joker once said….

Case Study #1: The Man Who Became 656 People

Johan Röhr is my personal hero at this point. This Swedish legend created over 650 fake Spotify artists. All by himself, and totaled over 15 billion streams.

He made chill-hop, study beats, and relaxation music…the kind of stuff you put on in the background to look intense.

Genius strategy, really.

I mean, think about it, when was the last time you looked at the artist who created some random white noise beat? Nobody actively listens to those things, so nobody’s going to notice that it’s all coming from the same guy.

Spotify, bless their algorithmically-driven hearts and annoying ads, put his tracks into 144 of their official playlists. All under these different fake names.

In one playlist called “Stress Relief” (which had 1.45 million followers), 41 out of 270 songs were just Johan wearing different hats. He ranked #64 globally.

When he got exposed, his record label called him a “pioneer in the mood music genre.”

So the lesson here is that if you’re a fake who makes money while doing it, you will be called a politician. But if you’re doing it on Spotify, you’re a pioneer!

Also, if you get caught, just rebrand it as innovation (or masterstroke (iykyk)). Works every single time.

Case Study #2: The Influencer Ponzi Scheme

Let’s talk about influencers.

Some brilliant souls with a lot of time on their hands actually went and analysed 100,000 Instagram and TikTok accounts. Over 37% of their followers were fake.

In beauty influencing, which is the most lucrative niche, it’s over 50%.

The thing is, once you cross that magical 100,000 follower threshold on these platforms, brand deals go from “we’ll send you free stuff” to “here’s $5,000 per post in actual cash.”

The cost to buy 50,000 followers would be around $200. Sometimes even less, if you know where to look.

So, let me do that math on this:

• Investment: $200
• New brand deal rate: $5,000 per post
• Number of posts before you’ve broken even: Less than one

This is literally better ROI than crypto scams.

And the brands are spending an estimated $4.6 billion a year on this nonsense. Most of them know, or at least strongly suspect, fake audience participation. But nobody wants to admit they’ve been paying people to advertise to literal robots. That would make them look stupid!

So the influencers keep buying followers. Brands keep hiring influencers. Bots keep liking posts. The algorithm sees all that “engagement” and shows it to more people. Everyone wins!

Well, except the brands. And the real creators who are trying to make it honestly. And anyone who actually values truth or buys all these beauty products , trusting those 1K likes and comments.

But who really cares about them?

The Underground Economy

Now, let me talk about something beautiful: there’s now an entire underground economy for fake account creation.

Researchers tracked the price of SMS verification codes, and they found that the prices for these services spike around elections.

Think about that for a second. There’s enough demand to manipulate political discourse that it actually affects market pricing.

Supply and demand! Capitalism baby!

It’s a functioning market with price discovery and arbitrage opportunities and stuff. For fakeness.

Step 4: Target the Art World

If you really want to game the system effectively, become a “photographer” or “artist.” And by that I mean: absolutely do not actually become a photographer or artist.

The photography and art world is in complete ruin right now, thanks to AI, and it’s never been easier to exploit.

You can look at an Instagram photo and think “wow, that’s beautiful” without knowing literally anything about how it was made! Was it shot on a $5,000 camera after years of training? Or was it generated in 30 seconds using ChatGPT? (the image generation in GPT5.5 is insane btw)

You literally can’t tell the difference anymore. It’s impossible. Even the professionals can’t tell the difference at this point.

There was actually a case where a photographer submitted a real photo to an AI art competition as a joke, and it won. Then he got disqualified when he revealed it was real.

We’ve gotten so used to AI slop that real photography now looks suspicious. The fakes have become so common that authenticity is considered weird now.

Again, let’s go over the numbers on this one.

Traditional Artist costs:

• 4 years art school: $100,000+
• Years developing a unique style: Priceless
• Camera equipment: $5,000+
• Time per photo: Hours
• Photos per day: Maybe 50

You, An Intellectual faker:

• Midjourney subscription: $30/month
• Time to generate a photo: 30 seconds
• Photos per day: Literally thousands
• Art school: lol no

The wedding photographer charges $2,000 for a grand wedding. AI-generated wedding photos cost $20. Sure, the people don’t exist, and the wedding never happened, but they look so incredibly good.

An Italian photographer, Emanuele Boffa, mixed AI-generated images into his Instagram feed without labelling them as AI.

When people called him out, his defence was: “Well, I call myself a digital artist, not just a photographer, so…”

Technically true! Technically completely irrelevant! But it worked, and that’s where we are now.

Work too good? People assume it’s fake. Actually fake? People assume it’s real. The entire system is going through some weird inception-like phase.

Pro tip: Mix real and fake stuff together. Keep people guessing. When questioned, just act deeply offended that they’d doubt your “artistic process,” maybe even question their taste…

Step 5: Relax, Nobody Can Verify Anything

This is honestly the best part. The secret sauce.

People can’t verify a damnn thing.

LinkedIn removed 84 million fake accounts in the first half of 2025 alone. (Oh, did u think linkedin was all true and professional….My sweet summer child… LinkedIn was fake even before AI slop)

Stanford researchers found over 1,000 LinkedIn profiles with AI-generated faces working for real companies. Actual legitimate businesses. RingCentral had them. Some startups making shipping container homes had them. Completely fictional employees on their roster.

Nobody noticed until researchers specifically went looking for it.

Studies show that people find AI-generated faces more trustworthy than real human faces (Can’t blame them on that one). because AI averages everything out. It gets rid of all the asymmetries and the weird quirks and gives you this bland and generically attractive face. Our lizard brains see that perfectly symmetrical face and just go “yes, I trust this person with my life and career.”

Even Voice cloning takes 20 seconds of audio now, and deepfake videos are good enough that finance teams have wired $25 million because their “CFO” told them to on a video call. Multiple people were on that call, and all of them thought it was completely real.

Look, the point here is that when the cost of creating a convincing fake keeps dropping, the profit from doing it keeps rising. (I mean , come on, we are talking $25 million!!)

Now, for all those people who are double-minded about this whole internet fakery and fear some karmic retribution for all the lying and fakeness, well, let me tell you what happened to all the pros of the game.

Johan Röhr, the guy with 656 fake Spotify artists, ya, well, he is considered a “pioneer” now and ya, No penalties whatsoever.

Jos Avery, the fake Instagram photographer, admitted everything publicly. He still has followers and he is still posting content.

Emanuele Boffa, who mixed AI into real photos, changed his bio slightly. Moved on with his life.

That’s literally it. That’s the punishment. Maybe you can add “digital artist” to your bio. Maybe even write a Medium post about your “journey” and “lessons learned.” Maybe you do nothing at all. Who cares?

The platforms might remove your account. So you make another one. Takes about ten minutes. There is literally no internet police.

Step 6: Enjoy Watching Everyone Enable You

Now for my absolute favourite part: the audience.

Your victims. Your beloved followers who enable all of this.

They know this is all happening. All of it. The fake followers. The AI-generated art. The completely fake personas, everything. And they just keep scrolling anyway.

I mean, the alternative is genuinely exhausting. Would you rather verify everything? Like, do reverse image search on every single photo you see? Ask for birth certificates from all those Spotify artists?

Nobody has time or the energy or even the slightest desire for any of that.

We’ve all entered into this unspoken agreement. You promise to make content that’s entertaining enough or inspiring enough or useful enough that we don’t ask too many questions.

It’s sort of like professional wrestling(WWE and stuff). Everyone knows it’s fake, but nobody cares. We’re just here to see the Undertaker choke-slam Brock Lesnar.

Now, worst-case scenario, even if someone figures out you’re fake, what are they actually gonna do about it? Make a Twitter thread exposing you? Cool. You’ll get a day of drama, maybe trending if you’re lucky. Then everyone moves on to the next thing. This happens on a weekly basis on crypto Twitter!

Because at the end of the day, your audience is getting exactly what they wanted: content to scroll past while they’re sitting on the toilet. Does it really matter if you’re real?

The answer, both financially and socially, is no. It doesn’t.

Conclusion: Welcome to the Post-Truth Economy

So there you have it. Your complete and comprehensive guide to being a dog on the internet.

Is it ethical? No, definitely not.

Is it sustainable long-term? Probably not.

Will anyone actually stop you? Absolutely not.

You see , There’s an old New Yorker cartoon about how no one can know if you are a dog on the internet (The one in the banner). It was supposed to be about internet anonymity and about the freedom to reinvent yourself online.

But I think the context has completely changed.

Now I don’t think it’s about “nobody knowing that you’re a dog.” It’s more about “nobody cares if you’re a dog, as long as you’re engaging or profitable or useful for the algorithm”

We built this whole infrastructure of social media, content platforms, creator economies, etc., on the assumption that humans would use it to connect with other humans, and then we screwed it up by rewarding being fake.

So here we are.

Look, I know I should probably tell you not to do any of this. That authenticity actually matters, and we have a moral obligation to be genuine with each other.

And I genuinely believe all of that! (ok, well at least some of that!!!)

But the thing is, the internet didn’t fail us on this one. We were the ones who optimised it for the wrong thing.

So ya go ahead. Be a dog on the internet.

Probably the only thing holding you back is the belief that your authentic self isn’t profitable enough.

And honestly, it is 2026….Who cares anyway…

If you want, I can also give you:

• a darker ending (full Black Mirror energy)
• or a more hopeful one (if you want it to feel less brutal)

(wink wink)

How a £300 Million Company Got Taken Down by a Phone Call and More Such tales from the World of Cybersecurity

Listen… I have a Bachelor’s degree in cybersecurity.

Spent three years on it. Did some genuinely cool projects. Passed the exams. Learned just enough to understand how networks actually get broken into and why most companies are basically a couple of bad decisions away from a very bad week.

Then I graduated… and immediately wandered off.

I Got into AI. Then Web3. Then back to AI. Then a bit more AI. At this point it’s mostly AI, with occasional detours into whatever the internet is currently obsessed with. (We listen and we don’t judge.)

Meanwhile, the cybersecurity degree just sits there in the background, like that one houseplant I keep meaning to water. Not dead. Technically still alive. Occasionally useful when someone at a party asks what I studied.

But yeah… mostly decorative at this point.

I usually only come back to it when something genuinely unsettling shows up in the news… or when I’ve posted about AI three times in a row and need to remind LinkedIn that I have writing range.

Sometimes both at once.

And This is one of those times. Look, To be fair, I actually did the reading on this one, so you’re going to get some real information and not just vibes and rants.

So let’s talk cybersecurity!

It keeps Getting Worse

I know, I know , you’ve heard this one before.

Cyber attacks are increasing. They’re always increasing.
Every year, its the same headlines and the same vague panics.

Nothing really changes, and you’re already zoned out by the second sentence in those headlines.

But this time… something actually did shift.

No, not just the number of attacks.

The nature of it.

For a long time, getting hacked required a specific kind of attacker. Someone technical enough to find a vulnerability, patient enough to exploit it, and motivated enough to do it manually.

There was a skill floor. Breaking into a system took effort, expertise, and time. That didn’t make it rare, it was very common, but it did create a natural ceiling on the volume of attacks.

Well, that ceiling is, at some levels, shattered.

And I know u know where I am going with this…AI

AI just made it super easy for people to hack!

Rather than just sticking to creating hackable vibe-coded websites and applications, AI just went ahead and democratised hacking.

You no longer need to know what a SQL injection is. You don’t need to understand packet sniffing or write your own malware. You can ask gpt for detailed instructions, use Claude Code to build some basic tools for yourself and if you’re so inclined, subscribe to a phishing-as-a-service kit on a criminal forum for around $200 a month, point it at a target, and watch it create personalised, grammatically perfect, psychologically calibrated attack emails at scale.( Don’t ask me how i know this!)

And the people who were already skilled at these stuff are now operating at a speed and scale that didn’t exist a few years ago.

CrowdStrike’s 2026 Global Threat Report clocked the fastest recorded attacker “breakout time” (the gap between first access and moving laterally through a network) at 27 seconds. That is literally the same time a certain world leader needs to decide on his next big invasion plan. (and tweet about it)

The AI Problem

This is where I get personally implicated in what I’m writing about.

The same tech I’ve spent the last few years aggressively shoving in people’s faces is now, in parallel, aggressively making data security a lot messier than it used to be.

Every company is rushing to deploy AI agents, hook them up to everything from internal systems to their brand new toasters and giving it access to calendars and emails and database,etc.

By doing this, they’re also, whether they’ve thought about it or not, creating a brand new attack surface.

An AI agent with access to your email, your CRM, your file system, and your Slack is a remarkable productivity tool.

It’s also a much more remarkable target.

The WEF’s Global Cybersecurity Outlook 2026, surveyed over 800 executives and security leaders across 92 countries and found that 87% of respondents identified AI-related vulnerabilities as the fastest-growing cyber risk they’re dealing with right now

And the same report found that only 64% had any kind of processes in place to actually check the security of the AI tools before deploying them.

Which means more than a third of organisations are just plugging things in and hoping for the best.(not a great strategy, tbh)

And yeah, there’s something a little uncomfortable about all of this, especially if you’ve spent time arguing that AI adoption is a good thing.

Look, I still think AI adoption is net positive, but “net positive” doesn’t mean “we’ll deal with all the bad stuff later.”

The Human Problem.

So, M&S.

Marks & Spencer, the British retailer who sells sandwiches (Yup) and expensive-looking cloths, has also somehow become the most discussed cybersecurity case study in recent years.

They got hacked by a group called Scattered Spider. The attackers didn’t break through a firewall. They called the IT help desk, impersonated an employee, talked a human being through resetting credentials and changing the MFA(Multi-Factor Authentication) settings.

A Ransomware followed. Then £300 million in estimated damages.

And just to be clear , Scattered Spider didn’t come up with this techniquie.

This kind of thing has been around forever. Social engineering ,which is just a nicer way of saying “tricking people and getting info”, has been one of the most reliable ways to break into systems since before the internet was even a thing.

Even Kevin Mitnick, probably the most well-known hacker out there, used to say his biggest advantage wasn’t technical skill.

It was getting people to trust him.

What’s changed is that AI has made all of this easier to pull off… and way easier to scale.

In January 2026, A group called ShinyHunters used a voice phishing call to get SSO credentials from a single CarGurus employee. That one call led to 12.4 million customer records being compromised.( The polite word for “getting stolen”)

The same group claimed breaches at Match Group ( Company that owns Tinder, Hinge and the whole set of apps your situationship is on), Panera Bread, and Crunchbase around the same time.

These where no movie-style hacking….

It was just people making phone calls and Sending emails and sounding believable.

Phishing used to be easy to spot. It use to have weird formatting, bad grammar or typos and texts like “Dear Valued Customer.”, etc.

AI phishing doesn’t have these things.

It has your name…..your colleague’s name….maybe even your pet’s name. It references a project you mentioned last week. Matches your company’s email format perfectly.

And instead of saying “URGENT CLICK NOW,” it says something like, “We’d appreciate your prompt attention to this.”

Which is probably why it slips past all the filters built for the old, sloppy phising versions.

And the WEF’s 2026 report flagged cyber-enabled fraud as one of its top three concerns this year, with 73% of respondents saying they or someone they know had been personally hit by it in the past twelve months.

The Supply Chain Problem

So now we’ve got AI-assisted attacks happening at a scale that just didn’t exist a few years ago… and social engineering that’s hard to spot.

That alone should be enough to make people a little scared.

But there’s a third layer to this that’s honestly a bit more worse.

Supply chains.

No, not the “shipping containers and trucks” stuff, the software kind.

The whole messy interconnected web of third-party tools, SaaS platforms, OAuth logins, integrations, vendor access… all the things modern companies depend on for their day-to-day functioning.

These things are very convenient and efficient , but they also mean that your security is only as strong as the weakest tool you’ve plugged into your system.

In March 2026, Stryker Corporation had its systems wiped in real time.

Employees literally watched their machines go down while security teams scrambled to respond. Offices had to shut down.

Around the same time, breaches at Telus Digital, Cognizant, TriZetto and a few others follwed a similar patterns

attackers weren’t targeting companies directly anymore. They were going after shared platforms and integrations.

Hit one weak link, and suddenly you’ve got access to everyone connected to it, no matter how solid their own security is.

And then, just this week (third week of April 2026), Vercel disclosed something similar.

An attacker compromised a third-party AI tool an employee was using… used that to take over their Google Workspace account… and from there, got into internal systems.

Step by step. Nothing fancy. Just following the connections.

The sequence was: AI tool → employee account → internal infrastructure.

You can train every employee.
You can roll out MFA everywhere.
Zero-trust, dedicated security team, the whole data security checklist.

And then some vendor or tool that you use for one tiny workflow gets compromised… and your data gets breached anyway.

Because from an attacker’s point of view, why spend months trying to break into a well-defended system…when you can just go after one of the forty integrations quietly sitting there with access to everything?

They figured that out a while ago.

A lot of “defenders” are still catching up and writing policies for a problem that’s already moved on.

The Skills Gap

I’m going to say something slightly uncomfortable here.

I have a cybersecurity degree and I left to go work on AI and Web3. And I am not even remotely unique in doing that.

The industry is short somewhere between 3 and 4 million professionals globally. People are studying it, the problem is that anyone good enough to do the work is also good enough to go get paid considerably more somewhere else.

AI is the most obvious “somewhere else.”

The WEF’s 2026 report found that 54% of organisations said limited skills and knowledge was the main reason they couldn’t adopt AI-driven security tools. Budget wasn’t the main issue. The tools are there but the people who know how to use them aren’t.

So what you end up with is this weird imbalance where attack surfaces gets bigger every year with more devices, more integrations, more AI plugged into sensitive systems…

and the people who know how to defend it are over here getting excited about the new ChatGPT 5.378 model release and/or generating Ghibli-style images.(and worse, some are just sitting there and writing blogs on the attacks!)

Why Companies Keep Losing The Same Fight

There’s a repeating theme in how breaches actually happen that I find annoying every time I read about it.

It’s rarely a sophisticated zero-day attack exploiting an unknown vulnerability.

In January 2026, researchers found a publicly exposed database containing 149 million records sitting in a misconfigured cloud environment. Nearly 100GB. Just out there.

It was not hacked…it was Misconfigured. Nobody noticed it and The database was essentially waving at hackers, asking to be stolen.

IBM’s X-Force found that 56% of vulnerabilities tracked last year could be exploited with zero authentication. Their North America lead said it plainly: attackers “don’t need zero-days. They just need valid credentials and a little bit of patience.”

Credentials pulled from old data breaches. Passwords and stuff reused across work and personal accounts. Data bought off the dark web. Or just… handed over by a tired help desk employee on a random afternoon who genuinely thought they were helping a colleague.

That’s usually all it takes.

The attacks that actually work aren’t clever. They’re just patient.

And the companies that get hit aren’t usually the ones that made some catastrophic technical mistake. They’re the ones that assumed complexity equals security.

They Locked down the obvious systems but left a few side doors open. They probably treated cybersecurity like a compliance checklist instead of something that actually needs attention and consistently underestimated the human layer.

You see, you can buy tools…but you can’t buy good judgment at scale.

Or a simpler way to put it would be that you can install the world’s best alarm system, but If someone calls the front desk, says they’re from IT, and talks an employee into handing them the keys, the alarm system wont help you.

So What Actually Needs To Happen

Yeah, so, this started as a LinkedIn brownie points exercise. I’m not going to pretend otherwise.

But it’s also one of those things that genuinely annoys me, because none of this is surprising. Nothing here is new or mysterious. It’s just what happens when a bunch of very predictable decisions stack up over time.

There’s no breakthrough fix for any of this. It requires boring, consistent, unglamorous attention to things that have been sitting on the to-do list for years.

MFA that actually holds up against phishing. Hardware keys. Passkeys. Not SMS OTPs that get bypassed in a five-minute phone call. Basic identity verification before resetting credentials. Treating vendor access like it matters just as much as your own systems. Training people for what attacks actually look like now, not what they looked like five years ago.

And, honestly, just paying security people enough that they don’t leave.

None of this is unclear. Everyone in security already knows what needs to be done. It’s just… treated as optional….Right up until it isn’t.

Because companies don’t fix this stuff when things are going well.

They fix it the week after the breach.

The Part Where I Admit I Should Probably Come Back More Often

Coming back to this after a while feels… strange. You knew it was going to be bad. You just didn’t know how bad until you actually looked.

The security fundamentals haven’t changed much, I mean, humans are still the easiest way in, stolen credentials are still the most common front door, and someone patient enough to make ten phone calls will beat someone with fancy tools and no patience almost every time.

What’s changed is that all of this now runs at a whole new level of speed and scale.The stuff that used to take a skilled, motivated attacker hours or days can now be automated and pushed out to thousands of targets before lunch.

And the truth is, I don’t have a clean ending for this. It’s not a problem with a satisfying resolution. It’s more of an ongoing situation where the sensible thing is to pay attention, do the boring stuff consistently, and stop treating security as something you think about after the breach rather than before.

Anyway.

I’m going to post this on LinkedIn now.

And then go enable MFA on a few accounts I’ve been putting off.