Velocity is everything. But most people confuse being busy with moving fast. Real velocity is the rate at which you close the gap between where you are and where you want to be. It requires dedication, focus, and a ruthless commitment to leverage.

So how the fuck do you actually do it?

More hours of focused work is a start, but it’s not really the answer. Let’s map it out:

• Side Hustle - 2 hrs/day, 5 days/week = 10 hrs/week, 40 hrs/month (1x)

• Part Time - 4 hrs/day, 5 days/week = 20 hrs/week, 80 hrs/month (2x)

• Full Time - 8 hrs/day, 5 days/week = 40 hrs/week, 160 hrs/month (4x)

• Dedicated - 8 hrs/day, 6 days/week = 48 hrs/week, 192 hrs/month (4.8x)

• Hardcore - 12 hrs/day, 6 days/week = 72 hrs/week, 288 hrs/month (7.2x)

• Intenso - 16 hrs/day, 7 days/week = 112 hrs/week, 448 hrs/month (11.2x)

Even the most intense work schedule only gets you 11.2x more than someone putting in 2 hours a day. That’s the ceiling on effort alone.

But each hour is not equal. In the software world there is the idea of a 10x engineer - someone so skilled that one hour of their work equals ten hours from a normal employee. And 10x is an understatement. There is a level of skill where the gap isn’t just speed - it’s capability. Below a certain threshold, you simply cannot do what a 10x engineer can do, no matter how many hours you throw at it.

The pottery class parable makes this real: a ceramics teacher split the class in two. One group was graded on the quantity of pots they made. The other was graded on the quality of a single pot. At the end of the semester, the quantity group produced both more pots AND better pots. > Volume creates skill. Skill creates leverage. Leverage makes each hour worth more.

On a long enough time horizon, the speed at which you iterate is the only thing that matters. Your rate of improvement - independent of where you start - determines who wins. The team that gets better fastest beats everyone else because their rate of growth compounds.

So the question becomes: how do you get more from your time?

The Pursuit of Excellence

If you don’t measure it, you don’t really care.

There are two ways to get more from your time:

1. Only work on the highest leverage, most important task

2. Only work in the highest leverage way

1. Work on the Right Thing

Most people fill their day solving B+ problems because they’re easy and feel productive. Quick wins, small optimizations, busywork that looks like progress. Meanwhile the A+ problem - the one thing that would make everything else irrelevant - sits untouched.

Imagine you’re starting a business. You’ve identified a problem you think you can solve. Two scenarios:

Person 1: Spends the next week researching all potential competitors and building out a feature list for what they want to offer.

Person 2: Spends the next day calling everyone they think has that problem to find out if it’s real and if their idea of a solution is what people actually want.

At the end of week one, Person 1 is no closer to validating if the problem is real or if people want the solution. Person 2 hit that point after 24 hours. That gives them the rest of the week to move on to the next most important task.

Assuming a 5-day work week - Person 2 is 5x more productive right out of the gate. Not because they worked harder. Because they worked on the right thing.

This is the “Four Parties” trap. Imagine you’re throwing a party and you split your guest list across four different events. None of them hit critical mass. None of them take off. Any single one of them could have been great - but none of them will work unless you pour everything into one.

Commitment is the elimination of alternatives. The most focused person does the fewest things outside the thing they are focused on.

This same idea applies across all domains. Every business has three core components: lead generation, conversion, and delivery. At any given time, one of them is the bottleneck. Find that constraint and pour everything into removing it. The question isn’t “should I do something different?” - it’s “what is the one thing stopping more output?”

2. Work in the Right Way

Let’s stick with the calling scenario. You’re calling everyone you know who has this problem. After each call, you open an app and write down what you learned. At the end of the day, you spend another hour going through all that data, drawing out common patterns.

Writing takes 10 minutes per call + 1 hour of analysis at the end. Each call lasts 20 minutes. That gives you about 14 calls’ worth of data in one 8-hour day, plus your analysis.

But what if you got AI to transcribe your calls and automatically summarize the findings? Now you cut the 10-minute write-up per call and the hour of analysis. That’s 10 more calls you can fit into the same day. So 24 calls in 1 day. In 2 days you’re at 48 vs 28. In 3 days, 72 vs 42.

Now go further. If you can work out how to get what you need from a call in 10 minutes instead of 20, you’ve doubled those numbers again.

This is the compounding effect of leverage. Each layer multiplies the last. This is what high velocity actually looks like.

The same math applies to delegation:

• The work yourself - 200 hrs/month (1x)

• Managing one person - 20 hrs/month (10x)

• Managing a manager - 4 hrs/month (50x)

• Setting strategy only - 1 hr/month (200x)

The key belief to kill: “No one can do it like I can.” With proper training and time, others can match or exceed you. And even if they only get to 80% of your level, you’ve freed up 100% of your time to apply your skill where it matters most - on the A+ problem.

When you hire, measure people by their rate of improvement, not their starting performance. A fast learner at 60% will outperform a slow learner at 80% within months.

And here’s where this gets wild: the same leverage math now applies to AI agents. OpenAI’s Frontier team built a million-line product in five months with zero human-written code - estimating 10x the speed of doing it by hand. Three engineers, 1,500 pull requests, 3.5 PRs per engineer per day. The agents wrote the code, the tests, the CI, the documentation, the internal tooling, even the production dashboards.

It’s not a story about AI. It’s a story about management. The same principles that make a great engineering leader - build systems, set clear expectations, remove bottlenecks, get out of the way - are exactly what made their AI agents productive. The technology changed. The leadership problem didn’t.

Let’s Get Real

You know the principle now. But how do you actually do it? Let’s get into the tactics.

Finding the High-Leverage Work

Ask these questions honestly:

1. What is holding you back right now? Not what’s annoying you. What is the single constraint that, if removed, would unlock the most growth?

2. If you could change one thing instantly, what would it be? This reveals what you already know is the bottleneck but have been avoiding.

3. What would you work on if you could only work 2 hours this week? This forces prioritization. The thing you’d still do with almost no time is probably the thing that matters most.

4. Are you solving B+ problems because they feel productive? Quick wins feel good. They also keep you from the hard, ambiguous work that actually moves the needle.

5. If your metrics for something are at or above industry benchmarks, is it really your bottleneck? If not, leave it alone. It’s not what’s limiting you.

“More” beats “better” beats “new.”

The fundamental question every business owner needs to answer is: why can’t I do more of what’s already working? Do more of what works (lowest risk), then improve quality (moderate risk), then try new things (highest risk).

Increasing Your Leverage on Every Task

1. Measure how long everything takes. You cannot improve what you don’t track. Time yourself. Be honest. Most people are shocked at how much time goes to low-value activity.

2. Identify what breaks your flow. What stuff do you have to do to keep progressing that isn’t actually progressing? Context switching, admin, tool setup, searching for information - these are the silent killers of velocity.

3. Set aggressive timelines. An end-of-day organization moves 7x faster than an end-of-week organization, and 30x faster than end-of-month. Unreasonable deadlines create a forcing function. They make you cut the bullshit and do only what matters.

4. Simplify ruthlessly. Complexity is the child of lazy thinking. It’s easier to say all four things than to do the hard strategic work of deciding which one matters most. Fewer things to do means fewer meetings, fewer people needed, less coordination overhead. One objective is easy to communicate and easy for the team to remember.

5. Automate the non-core. Any repeatable task that doesn’t require your judgment is a candidate for automation or delegation. The question isn’t whether someone else can do it as well as you - it’s whether your time is better spent elsewhere.

6. Hire at iteration 9. When entering a new area, don’t start from scratch. Hire someone who’s already at iteration 9 and apply a faster feedback loop to beat them. This is how you leapfrog instead of crawl.

The Formula for High-Velocity Execution

1. Big public goal - Everyone knows what you’re trying to accomplish

2. Aggressive timeline - Unreasonable deadlines create a forcing function

3. Frequent check-ins - Regular accountability without micromanagement

4. Small team - Remove people who create communication overhead

5. Let them cook - Clear goal, clear objective, check in regularly, get out of the way

This formula works identically whether “them” is humans or AI agents. The Frontier team enforced strict architectural boundaries - then gave agents total freedom within those boundaries. They cared deeply about correctness and structure. Within that, they let the agents express solutions however they wanted.

Enforce boundaries centrally, allow autonomy locally. Same playbook whether you’re leading people or agents. Same principles, faster clock speed.

What to Watch Out For

The cost of change is real. Every change incurs roughly a 20% dip in performance. The team has to relearn systems, customers get confused, momentum is lost. If your guaranteed downside is 20% and your potential upside is only 10%, that’s a bad trade. Stop changing things that are working.

Growth is not linear. It happens in stair-steps. You optimize at your current capacity until you hit a ceiling. Then you invest in the next level - new hire, new system, new process. You suffer a temporary drop in efficiency. Then you stabilize at the new, higher level. Then repeat. Expect the dips. They’re part of the climb.

Sometimes you have to let the little fires burn. Trying to solve the same problem three ways at once just dilutes effort from the one approach you know will work but takes time. Patience isn’t passivity - it’s figuring out what to do in the meantime while the main bet plays out.

Think in Systems, Not Tweaks

This is where most people fuck it up - whether they’re managing a sales team, running a marketing department, leading a construction crew, or deploying AI agents.

When someone underperforms, bad managers do one of two things: they micromanage every task, or they say “try harder.” Good managers ask a different question: “What is this person missing - context, tools, clarity - that would let them succeed without me hovering?”

This isn’t just a management insight. It’s the fundamental velocity insight. OpenAI recently ran an experiment where three engineers built a million-line software product in five months using AI agents - zero human-written code. When the agents failed, the fix was almost never “try harder” or “tweak the instructions.” Every time they hit a wall, the engineer stepped back and asked: “What capability is missing? What context is the agent lacking? How do I make it so this succeeds without me in the loop?”

That’s systems thinking. It applies whether you’re managing a sales team, a warehouse, a marketing agency, or a fleet of AI agents.

This is the same thing that separates a gym owner doing $30K/month from one doing $300K/month. The $30K owner is personally training clients, personally handling every complaint, personally writing every social post. The $300K owner built the onboarding checklist, the client escalation process, the content calendar with templates - then got out of the way. Same business. Different operating system.

The only fundamentally scarce resource is your attention. Everything else can be multiplied. Your time cannot.

The trap is thinking you can get velocity by staying in the loop - approving every decision, reviewing every output, hovering over every task. You can’t. That makes YOU the bottleneck. The restaurant owner who insists on tasting every dish before it leaves the kitchen has capped their business at however many dishes they can personally taste.

So instead of optimizing your involvement, optimize your systems:

• Encode your standards once, enforce them everywhere. Don’t review every output manually. Write it down as a checklist, a template, a rule, a quality standard. Your taste captured once and enforced continuously beats your judgment applied manually and inconsistently. This is how a franchise scales - the operations manual does the training, not the founder. This is how the best AI teams work - the guardrails do the reviewing, not the engineer.

• Give a map, not a manual. When you dump everything someone needs to know into one giant document, it fails. Too much context becomes no context. Instead, give a short overview that points to deeper sources of truth. This is how you onboard a new hire. This is how you brief a contractor. This is how you direct an AI agent. Short, clear, with signposts.

• Build feedback loops, not checkpoints. Don’t make yourself the checkpoint. Build systems where the person (or the tool) can see the consequences of their own work and self-correct. A sales rep who can see their own close rate dashboard doesn’t need you to tell them they’re slipping. An AI agent that can test its own output doesn’t need you to review every line. That’s not removing quality control. That’s making quality control automatic.

• Run cleanup continuously. Small problems compound into crises when left alone. The best operators fix small stuff daily so it never builds up. Weekly 15-minute process reviews beat quarterly fire drills every time. This is true for managing a team, running a warehouse, or maintaining a codebase.

The shift is from doing the work, to managing someone doing the work, to building systems that manage the work for you. At each level, your leverage multiplies. At each level, the instinct is to stay hands-on. That instinct is what keeps you slow.

The OpenAI team regularly had AI agents working on tasks for six hours straight - often while the humans were sleeping. But that principle isn’t unique to AI. The best business owners have teams that run while they’re asleep too. It’s the same game. Build the system right and you stop being the engine. You become the architect.

The Compounding Game

Here’s the thing about velocity that most people miss: it compounds.

Think of your career like a many-sided die. Each successful outcome turns more sides green. Red sides do nothing. The more you roll, the more you win. Eventually, hitting green becomes the rule rather than the exception. You don’t know how many sides your die has. You just keep rolling.

Outsized returns come at the 95th-99th percentile. You cannot reach the 95th percentile on five things simultaneously - you can only go that deep on one. The person who picks one thing and rolls the die a thousand times will crush the person who splits across five things and rolls each one two hundred times.

The world belongs to those who can keep doing without seeing the result of their doing.

Perfect timing is a myth. Perfect preparation is not. If your intention is to never stop, you can time everything perfectly - because you’ll always be ready.

The Frontier team’s throughput kept increasing - not because the AI got smarter, but because the systems got better. Each fix, each guardrail, each piece of encoded taste made the next thousand outputs better for free. That’s the compounding game in action. Build the system once. Let it enforce forever.

Whether you’re managing humans, AI agents, or both, velocity comes from the same place. Not from working harder. Not from tweaking inputs. From building systems that make the right output the default output - and then getting out of the way.

Focus relentlessly. Execute with volume. Measure obsessively. Build systems for leverage. Let it compound.

That’s velocity.

We’re told AI will either take all our jobs or, if it turns out to be a bubble, trigger an economic collapse.

Both possibilities are terrifying. But the real question isn’t whether AI works, it’s why our economic stability depends on it working at all. Where is the resilience? If one technology succeeds or fails, the vast majority of us should still be able to provide for our families and keep building a better future.

The fact that we’re even talking about a bubble points to something broken underneath. Not in AI itself, but in how several systems have collided to create a feedback loop we barely understand.

Two Systems

The first is how we value companies. A stock price isn’t a measure of what a company actually delivers, it’s a bet on expected future earnings. It reflects a story about value, not value itself. The explicit job of a public company CEO is to maximise that stock price, which means maximising the persuasiveness of that story.

The second is the attention economy. Attention drives economic value. The more attention something captures, the more capital flows toward it through investment, talent, and hype. Distraction isn’t a side effect of the modern economy. It is a growing sector of it.

Now watch what happens when these two systems meet.

The Collision

AI has captured our attention like no technology before it because it communicates in our own language. We’re not prepared for a technology that mimics our human method of communication. We see that in the words we use to describe what it’s doing: thinking, hallucinating, learning. That makes it uniquely compelling and uniquely attention-grabbing, because somewhere deep down we can’t fully differentiate between human and not human.

AI companies have used this to build narratives of extraordinary power: a coming utopia of abundance, or, if handled wrong, existential catastrophe. And they position themselves as the ones who’ll ensure we get the good outcome. Every startup founder believes passionately in their vision and works to persuade us to invest in it. That’s normal. But AI’s ability to seize attention means these narratives get amplified far beyond what the underlying value justifies.

You can see this happening in real time. When Nvidia reported earnings in May 2024, its stock surged by over $200 billion in a single day, more than the entire market capitalisation of most Fortune 500 companies. That wasn’t a response to Nvidia suddenly producing more chips. It was a response to a story about future demand. When that story wobbled a few months later on fears that AI spending might not pay off, hundreds of billions evaporated just as fast. Nothing had changed about what Nvidia actually made. Only the narrative shifted.

Or look at what happened when companies simply associated themselves with AI. When Buzzfeed announced it would use AI to generate content in early 2023, its stock doubled overnight. A company that had been struggling for years was suddenly worth twice as much because it attached itself to the right story. The underlying business hadn’t changed. The attention had.

This is where the fragility comes from. We haven’t gone all in on a technology. We’ve gone all in on a narrative.

The Loop

But the problem goes deeper than inflated valuations. There’s a feedback loop at work, and it’s self-reinforcing.

The attention economy doesn’t just distort how we invest. It erodes our ability to judge what’s valuable in the first place. When your every waking moment is filled (TV to fall asleep, podcasts while walking the dog, TikTok in every idle second, AI for every passing question) you lose the capacity to think clearly about what matters.

I see this in myself. I struggle to hear myself think. And if I’m experiencing this, so are millions of others.

A society that can’t judge value clearly invests capital into narratives rather than substance. Companies learn that capturing attention is more profitable than delivering real value, so they optimize for attention and metrics that support the narrative. That creates more jobs in the distraction industry: designing feeds, engineering engagement, manufacturing hype. More people working on distraction means more distraction in the world, which further erodes our collective judgement. Which drives more capital into narrative. Which creates more distraction work.

Eroded judgement → capital chases narrative → companies optimize for attention → more distraction jobs → deeper erosion of judgement → repeat.

This is the loop. And we’re feeding it with our money, our labour, and our attention.

Consider the trajectory of Instagram. It began as a photography app, a genuine creative tool. Today, its own internal research showed it was harming teenage mental health, and it buried the findings. Why? Because the features causing harm were the same features driving engagement. The product evolved not toward what was good for users, but toward what captured the most attention. That’s not a failure of Instagram. That’s the loop working exactly as designed.

Why the Economy Is Fragile

This loop explains the volatility. An economy where capital allocation is driven by attention rather than value has nothing solid underneath it. Stock prices swing on narrative shifts, not on changes in what companies actually produce. A single earnings call, a viral moment, a shift in hype can move billions. That’s not a market responding to reality, it’s a market responding to stories about reality.

Meanwhile, the work that genuinely keeps civilization running (farming, medicine, sanitation, education, infrastructure) is structurally undervalued because it doesn’t generate attention. A nurse keeping someone alive earns a fraction of what an engagement engineer at a social media company makes. The people doing the most essential work are rewarded the least, while the people engineering your next scroll are rewarded the most. The incentives are inverted.

Breaking the Loop

So what do we do?

The individual starting point is straightforward: reclaim your attention. Put the phone down. Sit with boredom. Think before you search. Let a question live in your head long enough to actually reason through it. These are small acts, but they’re acts of resistance against a system designed to prevent exactly this kind of thinking.

But individual discipline alone won’t break a structural loop. It’s like telling someone with an addiction to just stop. And make no mistake, these attention systems are engineered to be addictive. Meta, Google, TikTok, and Snapchat were all found to be designing their products to create dependency, as established in recent US court proceedings. TikTok and Snap settled out of court, presumably because they knew they would lose.

The research on addiction is clear: self-discipline alone is not enough. You have to change the environment. The same applies here. We need regulation that treats addictive design patterns the way we treat other public health risks. Not banning technology, but requiring companies to stop deliberately engineering compulsive use. The EU’s Digital Services Act is a starting point, but it focuses on content moderation, not on the attention-harvesting mechanics themselves. We need rules aimed at the loop itself: algorithmic transparency, limits on exploitative design patterns, and accountability when platforms knowingly cause harm.

But regulation alone takes time, and we can’t afford to wait. So alongside structural change we should each confront a more immediate question: what kind of work are you doing?

Is your work built on capturing attention without delivering real value? If so, you’re feeding the loop. That’s worth confronting honestly, not as moral judgement, but as a practical recognition that this work makes all of us, including you, more fragile.

Is your work keeping civilization running today? Waste disposal, healthcare, farming, justice, teaching, cleaning: without this, nothing else is possible. This is the foundation, and it deserves far more recognition and reward than it gets.

Is your work creating a better future? Scientists, engineers, artists, writers, builders, entrepreneurs solving genuine problems: this is human creativity at its best. Many paths here lead to dead ends. That’s fine. What matters is that the work is aimed at something real.

At the core of all three questions is a deeper one: what are your values? If you don’t know, if you’re just chasing economic success without examining what you’re building, you will almost certainly end up feeding the distraction industry. It’s where the money flows when no one is paying attention to where the money flows.

The Choice

The distraction loop won’t break itself. It’s too profitable, too self-reinforcing, too well-engineered. It breaks when enough people reclaim their capacity to think, judge, and choose, and then direct that capacity toward work that actually matters.

Figure out your values. Pursue them relentlessly. The future we leave our kids depends on it.