Field notes for the abundance age

These are the things I keep coming back to when I think about what to build next. The notes I’d want in front of me if I were planning ahead, whether I ran a two-person startup, a scale-up, or a big company.

They’re not gospel. They’re working beliefs, and I’ll keep updating them as the ground keeps moving. If you’ve read Two paths to the future, think of these as the practical notes from the same trip.

The cost of software is heading toward zero

You can build almost anything now for almost nothing. The constraint used to be capability, then it was time, and soon it’s neither. When making things costs next to nothing, the hard question stops being can we build it and becomes is it worth building at all. a16z make a version of this point well in their notes on AI apps in 2026, and it’s the question that actually matters now.

Taste is the scarce skill

When anyone can generate infinite output, the bottleneck stops being production and becomes judgment. Knowing what’s good, what to cut, what’s worth making at all, that’s the part the machines don’t hand you for free. AI gives everyone the same raw capability, so the edge moves to the people with a point of view and the taste to use it. Curation beats generation.

You get far fast, but 80/20 still rules

A working demo now lands in a weekend. That part feels like magic. The trap is thinking the demo is the work. The last 20%, the edge cases, the polish, the thing you can actually trust in front of a customer, still takes 80% of the time. The starting line moved. The finish line didn’t.

Build, ship, correct: don’t plan, build, then find out

When deploying is cheap, the smart move is to get something real in front of people and fix it in the open. A long planning phase followed by a big build followed by first contact with reality is the expensive way to be wrong. Be wrong faster, in smaller pieces, where it’s cheap to correct. Anthropic took Claude Cowork from first idea to release in about a week and a half, built largely by Claude itself, then iterated in public from there. That’s the whole build-in-public movement in one move: ship early, learn in the open, improve where everyone can see.

Start with distribution, not the product

Most ideas don’t mean much until you’ve worked out how you’ll reach the people who’ll pay for them. When anyone can build the product, the product stops being the hard part and distribution becomes the real moat. So start there. Build an audience, a community, a list, a reputation, and then build the thing you’ll sell into it. A decent product with a way to reach people beats a great product nobody can find.

No one knows the future

It’s more unpredictable than it has ever been. Anyone who tells you exactly how the next three years play out is guessing with confidence. So don’t build plans you can’t change. Build plans that are cheap to throw away. Treat your roadmap as a hypothesis, not a promise.

The most valuable companies are young, and the next giants may be tiny

Nvidia became the first four-trillion-dollar company in 2025, and a lot of the world’s most valuable companies didn’t exist a generation ago. The next step is stranger: the first one-person billion-dollar company is now openly expected. Size is no longer a moat, and being small is no longer a ceiling. So here’s the uncomfortable question to sit with: could your next huge competitor be a one-person business?

Your brand matters more, not less

As everything around it becomes a commodity, the one thing that can’t be copied is what you stand for. Your reputation, your point of view, who you actually are. That has to be solid and lasting, but also able to flex into new situations. LEGO is the obvious example: one core idea, creative play, held steady for decades while it reinvented almost everything around it, from sets to films to games, and even pulled itself back from near-bankruptcy without ever losing what it was.

There’s no reason to stay in your lane

When you can build almost anything, the category you started in stops being a fence. Apple is becoming a health company. Midjourney went from generating images to building a full-body ultrasound scanner. Amazon turned a bookshop into the infrastructure half the internet runs on. If your brand is strong and your capability is cheap, the lane is a choice, not a constraint.

The world is getting more global, but trends stay local

Distribution is now planetary. You can reach the whole world from a laptop. But taste, timing and context are stubbornly local. What’s obvious in Stockholm can be a non-starter in São Paulo. Build for global reach, but don’t assume one trend, one tone or one moment travels everywhere at once.

Physical AI is the next step

The next wave isn’t text on a screen, it’s perception, reasoning and action in the real world. Jensen Huang calls the moment near, the “ChatGPT moment” for robotics. Combining AI with physical AI lets you do things that were simply impossible before, at a cost that used to be unthinkable.

Bolting AI on gets you a few percent. Rebuilding around it gets you a multiple

Add AI to your existing org chart and you’ll get a nice efficiency bump. Rebuild the org and the process around what AI and physical AI can now do, and the gains are a different order of magnitude. McKinsey’s 2025 state of AI report points the same way: only a small share of companies see real bottom-line impact, and they’re the ones who redesigned how the work happens rather than sprinkling AI on top.

So don’t just pave the cow path. Before you automate a step, ask why it exists at all, because half the time it was a workaround for a limitation that no longer applies. And automate last, not first: understand the process, redesign it, run it manually until it actually works, then automate it. Automating something you don’t yet understand just makes a bad process run faster.

Data still matters, but simulation is catching up

Real-world data has been the great moat. That’s softening. When you can run millions of simulations for almost nothing, the value of slowly gathered real-world data starts to fall. And the world models that power those simulations are getting better fast, generating environments rich, consistent and physically plausible enough to train on in place of reality. The frontier is increasingly about how good your simulation is, not just how much real data you’ve hoarded.

The human is often the slowest part, but people need to talk more, not less

When the machines move this fast, the human in the loop becomes the bottleneck. And yet people need to talk to each other more than ever, precisely because the speed is so high that misalignment compounds quickly.

Slides were coordination, not the work

PowerPoint mattered when getting anything done meant pushing a plan up and down through layers of people. The deck was the interface between those layers, the way a plan survived the trip across the org chart. That world is becoming legacy. When planning can happen in an afternoon and the layers can be stripped out, a polished deck shouldn’t be mistaken for real work. The real work is the thing itself. Plenty of big orgs know this and are still slow to change, because the layers are where a lot of roles and comfort live.

Fewer people get more done

A small team in one room beats a big group spread across an org chart, almost every time. Most of what slows a large company down isn’t the work, it’s the coordination around the work: the meetings, the status updates, the endless back-and-forth between departments. Strip the layers out and a handful of people with one clear goal can decide at the speed of AI instead of the speed of the calendar. Every person you add buys a little capacity and a lot of communication overhead.

Friction is a feature when everything else is instant

When shipping is free and everything happens automatically, the value moves to where you deliberately slow down. Automate the reversible. Pause on the one-way doors: hiring, big commitments, trust, relationships, the calls you can’t take back. A little friction in the right place isn’t a bug to optimise away, it’s the moment you stop and use your judgment while the rest of the machine waits.

Robotics is expensive today and won’t be for long

The price of a robot today makes a lot of ideas look uneconomic. That’s temporary. Goldman Sachs already cut its humanoid cost estimate by around 40% in a single year. The only real floor is the price of atoms. Everything else, the compute, the software, the design, trends down over time. Plan for the world where robots are cheap, not the one where they’re a luxury.

Today’s luxury is tomorrow’s everyday

Ask what’s premium or expensive right now, then ask how it gets delivered to everyone on earth for almost nothing. The Apple Watch quietly put medical-grade heart monitoring on millions of wrists. DNA sequencing fell from around a hundred million dollars to a few hundred in two decades. Both would have been science fiction not long ago. The luxury of today is the default of tomorrow.

When building and running things gets cheap, bring more in-house

The old build-versus-buy maths flips. When you can stand up your own tools cheaply, it often makes more sense to own the thing than to rent it. Build your own sales system, shaped exactly to your business, instead of bending yourself around someone else’s. Own your robots instead of outsourcing your production. Vertical integration gets cheaper and more sensible the cheaper building becomes.

What stays scarce when everything is abundant

When building gets cheap and output gets infinite, value flows to whatever stays finite. Attention. Trust. Real human connection. A genuine original. The few things that can’t be mass-produced become the premium, and they’re mostly human, not technical. In an age of abundance, the scarce stuff is what people will pay the most for, so build around it rather than against it.

Make the world more fun

This is the one I’m most sure of. Delivering a product or a service isn’t enough anymore, because soon everyone can deliver the product or the service. The companies that make people smile, that make things a little more joyful and a little more human, are the ones that will be worth being around. Make the world more fun and people will choose you.