The Thinking Bottleneck: Why AI Made Your Notes the Most Valuable Thing You Own

A post hit the top of Hacker News this week that said something founders should pay attention to.

Stavros Korokithakis, a builder who's shipped multiple production projects using AI coding tools, wrote about his workflow. The part that matters: "I no longer need to know how to write code correctly at all, but it's now massively more important to understand how to architect a system correctly."

He's not alone. 356 upvotes and 308 comments, mostly from builders saying some version of the same thing: the bottleneck moved.

For years, the constraint on what founders could build was execution speed. How fast can you code it. How fast can you design it. How fast can you ship it. Those constraints are collapsing. AI writes solid production code. It generates designs. It handles the tedious stuff that used to eat your Wednesday.

So what's left?

Thinking.

The quality of your decisions. The clarity of your direction. The ability to spot what matters in a pile of noise. That's the new bottleneck, and it's not going away.

Execution Got Cheap. Direction Didn't.

Here's what Stavros describes about his actual workflow: he spends up to 30 minutes talking through architecture with an AI before a single line of code gets written. Planning the approach. Asking questions. Refining tradeoffs.

Then a cheaper model writes the code. Then different models review it.

The expensive part, the part that determines whether the project works or becomes a mess, is the thinking at the top. The human part. On projects where he deeply understood the domain, the code stayed clean at tens of thousands of lines. On projects where he didn't, "the code quickly becomes a mess of bad choices."

Same AI. Same tools. The difference was the quality of the thinking directing them.

This pattern shows up everywhere right now. Founders using AI coding tools get wildly different results depending on how well they can articulate what they want, why they want it, and what tradeoffs they're willing to make.

The question founders should be asking themselves: if thinking quality is the bottleneck, what are you doing to improve yours?

Thinking Is a Practice

There's a persistent myth that some people are just "clear thinkers" and others aren't. Genetics, maybe. Personality type. You either have it or you don't.

This is wrong, and we have decades of evidence for why.

Writing is thinking made visible. When you write down what you're working on, what you're worried about, what decision you're stuck on, you're doing something your brain can't do internally: you're externalizing your working memory so you can actually examine it.

Psychologist James Pennebaker at the University of Texas has studied expressive writing for over 40 years. His consistent finding: people who write regularly about their experiences show measurably better problem-solving, reduced cognitive load, and (here's the kicker for founders) better decision-making under stress.

More recently, a 2024 study in the British Journal of Health Psychology found that structured self-reflection practices improved goal attainment by 42% compared to control groups. The mechanism is simple: writing forces you to clarify what you actually think, which makes your next action sharper.

Founders who journal, who capture their thinking consistently, aren't doing it because it's relaxing (it's usually not). They're doing it because it makes them better at the one thing that still matters: making good decisions fast.

The Compounding Effect Nobody Tracks

Here's where it gets interesting.

A single journal entry is fine. Useful, maybe. A week of entries starts to show you patterns. Three months of entries is a strategic asset.

Why? Because you start to see what you were thinking about before that product launch, before that hiring decision, before that pivot. You get receipts on your own judgment. You can check: was I right about that thing I was worried about in January? Did the thing I was excited about in February actually matter?

Most founders rely on memory for this. Memory is terrible at it. Memory is lossy, biased toward whatever happened recently, and weirdly selective about what it keeps. Your current note-taking system probably doesn't help either, because it's optimized for capturing information, not for showing you your own thinking over time.

The founders who compound the fastest are the ones who can say, "I noticed this pattern 3 months ago, wrote about it, and now I'm seeing it confirmed." That's a real edge. It's the kind of thing that helps you build in public with genuine signal instead of performative updates.

What Good Founder Thinking Capture Looks Like

Let's be specific. Because "write more" is not actionable advice.

1. Capture decisions, not just tasks.

Most note-taking is task management in disguise. "Build the onboarding flow. Fix the billing bug. Call the investor." That's a to-do list wearing a disguise.

The entry that compounds is: "Decided to prioritize onboarding over billing because 3 users this week mentioned confusion in the first 5 minutes. Billing bugs affect 2% of users. First impression affects 100%." Now you have the reasoning. Now future-you can audit whether that was smart.

2. Write the thing you're avoiding.

If there's a decision you keep pushing to next week, that's exactly the one to write about. The goal is describing it clearly, nothing more. "I keep avoiding the pricing conversation because I'm scared we're undercharging but also scared that raising prices will kill our early momentum."

Writing that down doesn't solve it. But it strips the vagueness away. You can actually work with "scared of losing early momentum" in a way you can't work with a general sense of dread.

3. Record what you notice, even when it seems small.

"3 people this week mentioned they use Compass right after their morning standup." That's a throwaway observation. Useless in isolation. But 6 weeks later, when you're trying to figure out your ideal activation window, it's gold.

The best founder notes are half observation, half "I'm not sure why this matters yet." That uncertainty is the point. You're building a dataset of your own noticing.

4. Review weekly.

Capture without review is just hoarding. Once a week, spend 15 minutes reading what you wrote. Circle the things that still feel true. Cross out the things that don't. Write a 3-sentence summary of what you learned this week.

This is where the compounding actually happens. The review turns raw notes into refined thinking. It's the difference between a pile of ingredients and a meal.

Why AI Makes This More Urgent, Not Less

There's a tempting logic that goes: AI handles the execution, so I can think less. I just need to point it in roughly the right direction.

The opposite is true.

When execution was slow and expensive, a mediocre decision still took weeks to play out. You had time to course-correct. The latency between "choose direction" and "see results" was long enough to mask fuzzy thinking.

Now? AI collapses that cycle. You can go from idea to shipped feature in a day. Which means a bad decision shows up in production by Thursday instead of next month. The faster you can ship, the more your thinking quality matters, because you're making more consequential decisions per week.

Stavros's workflow is proof of this. He says the planning phase with an AI architect takes the most time, "sometimes even up to half an hour of back-and-forth until we finalize all the goals, limitations, and tradeoffs." The writing-of-code part is the easy bit. The thinking part is what he invests in.

And the people getting bad results? "On projects where I have no understanding of the underlying technology, the code still quickly becomes a mess of bad choices." Same tools. Different thinking quality. Different outcomes.

The Asymmetry Founders Miss

Most founders spend 80% of their time on execution and 20% on thinking. The tools now suggest the ratio should be closer to inverted.

If AI handles execution, your 80% should be: talking to users, writing about what you're learning, reviewing your past decisions, and getting clearer on where you're going. The 20% is pointing AI in the right direction and reviewing the output.

The founders who figure this out first are going to look like they have superpowers. They won't be faster typers or better prompters. They'll be clearer thinkers. And that clarity will come from one boring, unsexy practice: writing things down, consistently, and actually reading what they wrote.

You don't need a productivity system. You don't need 17 different apps crammed together. You need one place where you capture your thinking, and a habit of reviewing it.

Who You Become

The Reflective Founder doesn't journal because a blog post told them to. They journal because they've seen what happens when they don't: reactive decisions, repeated mistakes, and the vague sense that they're running fast in no particular direction.

They're the founder who, in a board meeting, says "I wrote about this exact scenario 3 months ago, here's what I thought then, and here's what actually happened." They're the founder whose intuition gets sharper over time because they're not relying on intuition alone; they're cross-referencing it with their own written record.

The ones who take their own thinking seriously. That's who wins when execution is free.

Compass was built for this. One place to capture everything (voice, text, whatever form the thought arrives in), with the context and patterns that only emerge over time. But the tool only works if you do the practice. The practice is what compounds.

Your notes are now your most valuable startup asset. AI writes the code. AI handles the design. Your thinking is what's left, and it's everything. Captured, reviewed, compounded.

Start today. Open whatever you've got, even if it's a blank doc, and write down the 3 biggest decisions sitting in front of you right now. Write why they're hard. Write what you're leaning toward and why.

Then come back next week and see if you were right.

That's the practice. Everything else follows.