If there's one thing that depreciates faster than LLMs, university is definitely high on the list. University is also a symbol of knowledge. Yet knowledge workers can more or less sense that their so-called 'knowledge' is far from future-proof, nowhere near the job security and lasting relevance of blue-collar work.

AI has played a role in the movement of ‘knowledge democratisation’, not as the root cause, but as a catalyst. Knowing things and being able to recite them is no longer cool or useful on its own. Without the moat of knowledge, what do knowledge workers have left? Can they survive the rising tide of AI?

Buffett was definitely right about the tide revealing who’s been swimming naked. But just as importantly, the tide also washes away surface-level dirt that was never going to stick, uncovering the enduring structures beneath that create opportunities in different forms across different eras. The best move isn’t to become just another layer of dirt waiting to be washed away, but to understand what makes something truly stick. As the new tide surges forward, a new playbook for knowledge workers is being written.

Developing the Intuition for Good Ideas

Most startup people claim that ideas don’t matter as if it’s an indisputable fact. But really, it’s just an anecdotal insight born from watching too many ‘idea people’ who fall short on execution.

Idea matters. When an idea is big enough, it becomes a vision. And I think we can all agree, at least in hindsight, that we know what a good vision looks like and tell it apart from a bad one. The source of a good idea often has more to do with the person behind it than the idea itself. Winning ideas tend to emerge from a cluster of potentially good ones, shaped by the paths that person has explored or contemplated.

Becoming someone who consistently generates good ideas matters. When it becomes part of who you are, the process of idea generation turns intuitive, and this is what Daniel Kahneman would call System 1 thinking.

When Jeff Bezos left his job in New York for something called the Internet after seeing its 2,300% year-on-year growth, BYD entered the petrol car market in 2003 as a battery maker just as China joined the WTO, Lei Jun founded Xiaomi with ‘affordability’ in smartphone devices at its core, they saw things most people only saw decades later.

For every good idea, there are probably hundreds of bad ones. A common source of bad ideas is false knowledge, some of which even becomes consensus. In idea formation, consensus can be poison. The real world doesn’t have much incentive to deliver truth to individuals. What we often end up with is a world full of consensus shaped by the interests of profit-seeking groups and regulatory capture.

We all nod when we hear that being ‘contrarian and right’ is a powerful combo, but the problem is that ‘right’ is something we can usually only know after the fact. For the sake of action, it may be more helpful to aim for ‘contrarian, deep, and lots of them.’ We can end this part with answers to Peter Thiel’s iconic question, start with one, and then lots of them:

Creating the Knowledge Monopoly

The value of knowledge to an individual gets diluted as that knowledge spreads. This mirrors the Law of Shitty Clickthroughs in growth marketing:

Once knowledge becomes widespread, it loses its edge. Scarcity creates value, not just in commodities, but in ideas too. The mechanism of decay varies. Once knowledge becomes common sense, the willingness to pay for it evaporates.

The counter to the Law of Shitty Clickthroughs in growth marketing is to give up on scalability and focus on niche channels with limited reach and short lifespan. Similarly, the way to address the declining value of knowledge is to create a monopoly on it.

Contrary to popular belief, not all knowledge is freely or fully available even in the era of AI dominance. Sure, surface-level answers are everywhere. But in the real world, no two environments are exactly alike, and countless unwritten rules exist beneath the surface. This opens the door to context-rich, off-menu expertise, something AI cannot easily replicate or dilute. In the right conditions, people can build a knowledge monopoly. That’s where the real value lives.

People should think about this the same way they think about product growth. Once core value is created, the product, it needs the right audience to realise its worth. For individuals, the most sensible move may not be to go narrow, but rather to get yourself out there and let the right audience find you. This could explain the empirical successes we see in building a public personal brand.

Executing Second-layer Moves

When ChatGPT first launched, many assumed coding jobs would disappear given AI’s ability to write code. Over time, AI has indeed taken over more of that task, rising from writing 50 percent of code to 95 percent today. In the near future, it might write 99 percent or even 99.9 percent of code for humans.

But the vast majority of people will still outsource this task to someone else, as long as coding does not become as easy and intuitive as other office tasks. We have seen this pattern repeat itself in many domains. In investment management, factor investing and asset allocation have become common sense. Even after Meb Faber famously said that asset allocation does not really matter as long as you include the right core assets, most of his clients still delegate the allocation task to him, not because they do not understand it, but because they do not want to deal with it.

This is what I call second layer moves. These are actions that are entirely doable but come with just enough hassle to stop most people. You could do it yourself. The knowledge is out there. The tools are available. But the small frictions in the setup, the mental load, the unknowns are just enough to hold most people back. That is when they turn to the person who makes it look effortless.

Most people stop at the first layer, consuming content, saving bookmarks, collecting advice. A smaller group moves to the second layer, shipping things, building systems, solving specific problems, connecting dots in the real world. The second layer isn’t that hard, but it does require breaking inertia and pushing through minor hassles. And that’s exactly what allows value to compound.

In a world flooded with knowledge, the edge isn’t in knowing more. The edge is moving while others are pausing. It doesn’t even have to be original; just doing what others aren’t willing to do is already creating value.

Think deep, develop the intuition for good ideas, create and be known for your own knowledge monopoly, and act when others stall. This is the new playbook in the AI era.