Rich people have had agents for a hundred years. They're called assistants.

So we don't have to guess what life with a perfect agent looks like. Look at what the rich do with theirs. The assistant renews the passport, calls the plumber, picks up the dry cleaning. The boss picks the house, the painting, the holiday.

Nobody with an assistant stopped shopping. They stopped doing errands.

That's what agents will do to the web. The pitch for AI browsers is that you'll never visit a website again. That pitch treats the web like a to-do list. Most of it isn't one.

Errands and choices

Some of what you do online is an errand. Reorder the printer ink. Find out when the pharmacy closes. Book the same flight as last time. Cancel a subscription. You already know what you want, and the website is in the way.

Agents should take all of it. Nobody will miss it.

Here's the rule. If you can say what you want in one sentence, an agent can get it. "The same ink as last time" is a sentence. "A sofa that feels right in this room" isn't. Agents win wherever one product is as good as another, and lose wherever it isn't.

Everything else is a choice, and you can't hand off a choice you haven't made yet.

An agent needs to be told what you want. For anything that matters, you don't know until you look. You open twenty flats and find out you care about light more than size. You scroll through beach holidays and realise you want the mountains. The looking is how the wanting gets made. An agent can only answer a question you already have.

That's why the most expensive shops in the world are built for people who could send someone else. They come in person anyway. Choosing is the part they kept.

And most of the time people spend online was never an errand to begin with. About three-quarters of all mobile data is video. Nobody sends an agent to watch something for them.

The obvious objection is that the agent will learn your taste and pick better than you. Maybe it will. A good assistant knows the boss's taste too, sometimes better than the boss does. The boss still picks the house. For the things you care about, choosing isn't a cost to cut. It's the thing you're buying. Nobody wants the optimal holiday. They want to have chosen it.

Silicon Valley's favourite demo

Every assistant launch ends with a booking. In 2018, Google's Duplex phoned a hair salon and booked a haircut. In 2024, Rabbit's R1 planned a holiday on stage, flights and hotel included. In 2025, OpenAI's Operator booked a table for two. It's the demo because it's easy to show, not because anyone wants it.

"The same flight as last time" is an errand. "A flight to Lisbon in May" only looks like one. It's a stack of tradeoffs: price, times, layovers, the airline, the seat, the bags. You find out which ones you care about when you see them. Silicon Valley keeps pitching it. People keep booking their own.

Amazon already tried an agent that shops for you. It sold tens of millions of Alexa speakers partly on the idea that you'd shop by talking to them. In 2018, The Information reported that only about 2% of owners had bought anything by voice that year, and most who tried never did it again. A voice can reorder paper towels. It can't show you a jacket.

Your agent isn't your assistant

There's one more reason to keep choosing yourself. A rich person pays their assistant. Your agent will be free, and free things work for whoever pays for them. ChatGPT already shows ads.

We've seen this before. In 1976, American Airlines started putting its Sabre booking computer into travel agencies. It showed flights from every airline, and American's came first. Agents booked from the first screen, so they booked American. It took until 1984 for regulators to make the systems show an unbiased screen.

For an errand, a biased agent costs you almost nothing. Ink is ink. For a choice, the bias is everything. If the picks might be for sale, you'll want to see the options yourself.

What the web loses

The errand pages die. Recipe sites with a life story above the recipe. "Best laptops of 2026" lists. Pages nobody wanted to visit, built to catch you on the way to one fact and show you an ad. An agent takes the fact and leaves. It's already happening with plain AI search: when Google shows an AI summary, the top result gets 58% fewer clicks.

Good. Those pages were a tax on errands, paid in your time.

What's left is the web people visit on purpose. Fewer people arrive by accident, so every page has to be worth the trip, and design matters more than it used to, not less.

Everyone is about to get an assistant. Rich people already showed us what happens next. The errands disappear. The shopping doesn't.

Sources

• Mobile network traffic update (Ericsson Mobility Report)
• Update: AI Overviews Reduce Clicks by 58% (Ahrefs, February 2026)
• OpenAI to start testing ChatGPT ads across Free and Go tiers (SiliconANGLE, January 2026)
• The Information says Alexa struggles with voice commerce (Voicebot.ai, August 2018)
• Google's AI assistant can call and book appointments for you (ScienceAlert, May 2018)
• What exactly is the Rabbit R1? CES 2024's AI breakout hit explained (TechRadar, January 2024)
• Introducing Operator (OpenAI, January 2025)
• Sabre history (Sabre)
• Computer Reservations System (CRS) Regulations (US Department of Transportation, Federal Register, 1997)

In ten years, intelligence will be priced like electricity. You'll pay for it by the unit, it'll be cheap, and it'll be boring.

The interesting question isn't whether AI gets smarter. It will. The question is who owns it.

The price only goes one way

Every year, the same level of capability costs about a tenth of what it did the year before. A task that once needed the biggest model in a datacenter moves to a smaller model on rented hardware, and eventually to one small enough to run on your laptop.

Open weights trail the best closed models by months, not years. Whatever the best models can do today, you'll be able to download and run yourself not long after.

When something gets that cheap that fast, it stops being a product and becomes a commodity. Nobody brags about their electricity.

The model isn't the moat

If models commoditise, what's left?

• Compute. Intelligence is FLOPS turned into answers. Whoever has the cheapest FLOPS has the cheapest intelligence.
• Data. The kind you can't scrape: from your own users, your own sensors, your own mistakes.
• Taste. Knowing which of a thousand generated answers to keep.

A company whose only edge is a slightly better model is on a deadline: its lead lasts until an open model catches up. Unless catching up becomes illegal.

Doom sells

In May 2023, the heads of OpenAI, Google DeepMind and Anthropic signed a one-sentence statement: the risk of extinction from AI should be a global priority, alongside pandemics and nuclear war. The same month, OpenAI's CEO told the US Senate that models above a certain level of capability should need a government licence.

Then they all went back to building them.

Maybe they mean it. It doesn't matter. Look at what the fear buys.

It sells the product. Nobody pays a premium for a commodity, but a model that might end the world is the most important thing you could ever invest in. Doom is how you keep AI from being boring.

And it writes the rules. If AI is a weapon, only licensed labs should be allowed to build it, and the labs asking for licences are the ones who'd get them. Open weights become the dangerous thing, and the deadline goes away.

George Hotz made the opposite case when he debated Eliezer Yudkowsky in 2023. The defence against powerful AI isn't keeping it in a few hands. It's putting it in everyone's.

What I actually worry about

It isn't a model waking up with its own plans. It's that the few companies with the most compute end up running the thinking for everyone else, and everyone else rents it.

When you rent your thinking, someone else decides what it's allowed to think about, what it costs next year, and whether you get to keep using it. That's a lot of power in very few places, and nothing from science fiction has to happen for it to go badly.

The answer to that isn't rules that lock the current leaders in. It's making sure anyone can own and run their own models.

Own your compute

This is already more possible than people think. A laptop can transcribe a lecture and run a useful language model today. The speed limit is set by memory bandwidth, and bandwidth per dollar only goes up.

Every year, bigger models fit on your own machine. The future I want is one where that keeps happening until you don't need anyone else's.

Renting will stay cheaper. A datacenter reads the weights once and uses them for hundreds of people at the same time. Your laptop reads them for you alone. Power from the grid is cheaper than running your own generator, too.

That's fine. You don't have to run your own. You have to be able to. The option to leave is what keeps the rent fair.

If you can't run it, you don't own it.

What that means for me

If every layer of AI turns into a commodity, the people who matter are the ones who understand all of them — from the matrix multiply at the bottom to the product someone uses at the top.

So that's what I'm trying to learn. Not one framework. The whole stack, down far enough that when the tools change, I don't have to start over.

I might be wrong

Maybe the biggest models win outright and stay so far ahead that nothing you can own ever catches up. Then the future is a few enormous datacenters, and the rest of us are customers.

I don't think so. Every computing wave so far started centralised and ended up in your pocket. I'd bet on this one doing the same.

A boring future: intelligence as cheap as electricity, and a generator in every house.

Sources

• Statement on AI Risk (Center for AI Safety, May 2023)
• OpenAI CEO Sam Altman Agrees AI Must Be Regulated (TIME, May 2023)
• George Hotz vs Eliezer Yudkowsky AI Safety Debate (Dwarkesh Patel, August 2023)

I hate flat interfaces. Grey rectangles, one accent colour, thin icons, and every app looks like a settings page. Software used to feel like objects. Now it feels like forms.

Everyone says flat won because it was better. I think it won because it was cheaper.

Flat won on price

In 2013 Apple shipped iOS 7 and threw out the leather, the felt and the wood. Microsoft had already gone flat with Metro. Google followed with Material Design in 2014. The story was honesty: a screen is a screen, so stop pretending it's a notebook.

Here's the part the story leaves out. Every texture in the old style was painted by hand in Photoshop, exported again for every screen density, then wired in by an engineer. Every new device meant doing it again. That look took teams of designers and engineers years to build and keep alive. A flat rectangle is one line of CSS, and it scales to any screen for free.

Flat wasn't a philosophy. It was a budget, and the philosophy was written afterwards.

What we lost

The leather was bad. Nobody misses the stitching.

What we lost was information. A raised button told you it could be pressed. A shadow told you what was on top. Light came from somewhere, so your eye knew which way was up. Flat design threw out the information together with the decoration.

Now you hover over things to find out whether they're clickable. That's a step backwards, and we called it minimalism.

Detail just got cheap

The cost that killed skeuomorphism is going to zero.

A model can write the shader for a material, the lighting for a scene, and the CSS for a surface that catches the light. It renders at any size, so there's nothing to export. An image model paints a texture in seconds. The slow, expensive part of the old style is now the fast, cheap part.

What used to take a team of designers and engineers months now takes one person and a model.

When detail costs nothing, flat loses its only real argument.

Even Apple is moving back. In 2025 it introduced Liquid Glass: refraction, highlights, depth. Big companies don't change their whole design language for fun.

The catch

Ask a model for an interface today and you get flat, because it learned from ten years of flat. AI makes depth cheap, but it won't choose it for you. That part is still taste.

And skeuomorphism done cheaply is worse than flat. Fake glass on everything, gradients with no light source, shadows pointing in three directions. At that point you'd miss the grey rectangles.

The point was never decoration. It's physics. Things that look pressable are pressable. Materials behave like materials. Light comes from one place. If you can't say where the light comes from, don't add the shadow.

This site

This site has a folder that opens, a badge that hangs on a strap, and a pile of posts you deal out with a click. None of it is necessary. All of it is the point: you can tell what something is, and what it does, just by looking at it.

A few years ago, that much detail would have been months of work for a team. Building with AI is what made it possible for one person.

Or flat wins again

Maybe flat wins again, for the same reason it won the first time: it's easy to keep consistent. Fifty people shipping one product can all draw a grey rectangle the same way. Depth needs someone who owns the light.

But "easy to keep consistent" is not a good enough reason for software to have no soul.

The excuse was cost. The cost is gone.

Every local-model benchmark you've seen was capped by one division.

To write one token, a model reads every weight it has. All of them, once per token. An 8-billion-parameter model reads 8 billion numbers to produce one piece of a word, then does it again for the next.

The arithmetic on each weight is tiny. Moving it from memory to the chip is not. So the speed limit isn't how fast your chip computes. It's how fast it can read.

tokens per second ≤ memory bandwidth ÷ model size in bytes

Do the division

My laptop is an M1 Pro from 2021. Apple quotes 200 GB/s of memory bandwidth.

An 8B model quantised to 4 bits uses half a byte per parameter, so it's about 4 GB, a little more in practice.

200 ÷ 4 = 50 tokens a second. That's the ceiling. Real runs land below it, because nothing is perfectly efficient. They don't land above it.

Here's the same division for a few machines, with 4-bit weights. A dash means the model doesn't fit in memory at all.

The bandwidths are 200, 170, 614 and 1,792 GB/s. Everything else is division.

Notice the last row. The 5090 is about nine times faster than my laptop, and it can't run the model at all. Speed doesn't help if the model doesn't fit, and on graphics cards, memory is the expensive part.

Newer isn't faster

Look at the M6 column. It's Apple's newest chip, released in September 2026, and it reads memory at 170 GB/s. My five-year-old M1 Pro does 200. For running models, the new chip is slower.

The generation number barely matters here. The tier does. Pro and Max chips have wider memory buses, and that's the whole game: an M5 Pro reads at 307 GB/s, an M5 Max at 614. If you're buying a Mac to run models, pay for memory bandwidth, not for the newest chip name.

What falls out of this

Quantisation is a speed trick, not just a memory trick. Going from 16 bits to 4 makes the model a quarter of the size, so the ceiling goes up four times. That's why everyone runs 4-bit locally.

Your chip is mostly waiting. When you generate text for one person, the compute units sit idle between reads. That's also why a datacenter is cheaper per token than your laptop: it reads the weights once and uses them for hundreds of people at the same time. You're reading them for one.

Mixture-of-experts changes the maths. An MoE model only reads the experts each token is routed to, so its speed is set by the active parameters, not the total. You still need the memory for all of them.

Long context slows you down. Every new token also reads the attention cache for everything before it. At the start of a conversation that's nothing. Thirty thousand tokens in, it isn't.

Reading the prompt is a different problem. The prompt is processed in parallel, so that part is limited by compute, not bandwidth — and compute is the thing a laptop is actually short of. That's why pasting a long document feels slower than getting the answer.

I checked

I ran Qwen3 8B on my M1 Pro through Ollama. At 4 bits it's 5.2 GB, so the division says 200 ÷ 5.2 = 38 tokens a second, at most.

It did 22. Three runs: 22.1, 21.9 and 22.0.

That's 58% of the ceiling. Below it, never above it, exactly where the maths said it would be. I didn't need the benchmark to know roughly what it would say. I just needed a division.