You didn't miss my last newsletter, I only send when it's good enough and worth your time. Sometimes that takes two months."

The Big One.

I’m in two minds.

I always thought that fully self driving cars would be impossible. How would a car ever know the difference between a paper bag and a rock in the road, or get a sense of when a kid would chase a ball into the road.

It turns out with enough data and enough compute, rather like a person, you can (mostly?) solve that problem.

Some people in the world think this sort of applies to everything. With enough data and compute you can have humanoid robots that do any job, from laundry to mining to teaching to nursing, everything comes down to more data and more processing power.

They think with AI, you can invent new materials, develop new drugs, you can predict the ROI of a tagline, or pretest a new Kombucha variants success or answer the big questions like what’s best, Finish Quantum Powerball , or Ultimate Powerball, or Power Powerball.

This week, more than any other, which we didn’t think possible , all manner of craziness is afoot, Openclaw, 5.3 Codex, Claude Cowork, it’s relentless waves of “breakthoughts” and ever more dramatic claims. Are these things becoming emergent or are they just doing precisely what we precisely told them to do.

I’m in the godforsaken middle AGAIN.

I think 97% of people are not thinking about AI enough, or they underestimate it. They are a somewhat vulnerable, ignorant, unprepared for some true changes that will almost inevitably come. From being scammed by clones to sharing fake news, it will impact everyone.

And we’ve a painfully loud 3% of the world thinking about it far far too much. People mesmerized by promises, who underestimate the truly VAST real world challenges of turning even the most magical tech into something that can be implemented or add value, in the slightest of ways.

It’s all taking on the characteristics of religion.

Evangelicals saying the end of the world is nigh, exasperation over the non believers, the heretics who ask questions, it would be easier to take seriously if they weren’t the same people who barked like seals at the Metaverse, Web3 and how FIAT was going to fall to NFT’s. There’s a lot of Gel Mann Amnesia too, people who know least about real world jobs, with no clue that more people use fax machines that Slack. Who continue to think truck drivers jobs or Architects are at risk because they have no idea what they actually do.

If we look at real world implementation there’s all manner of shocking stats.

Everyone loathes MS Copilot, OpenAI seems to use remarkably little of their own AI, 95% of Gen AI pilots fail, 85% of AI projects show no significant impact, even the people who boasted most about integrating Vibe coding went a bit quiet after the IPO.

So it seems like OpenAI is struggling, Anthropic will say anything to get noticed, Microsoft have to talk up their own book, no wonder they keep promising the next release will change it all or that “we’re making something so good, we’ve had to make it less powerful”, it feels like a con job.

Right now we’re at the inevitable loggerheads of the algorithmic age, where only extreme opinions get noticed. The sides are talking past each other. Again “Have fun staying poor’ from the NFT era feels like the vibe.

For what it’s worth I remain very much bemused by both sides.

I think

• Something can be incredibly magical but not particularly valuable.

• Something can keep getting better, but a tiny error rate may make it worthless.\

• I don’t think the world is particularly rational in ANY way.

• Tech people are staggeringly odd and their thoughts and behaviors have no impact on 97% of the world.

• Companies are motivated by risk avoidance over efficiency

• The depth of integration which creates change, not the power of the technology.

• Something can change the world, but take a VERY long time to, like EVERYTHING EVER before.

At the heart of this are many interesting questions, but 2 big ones for now.

1) To what extent can every problem be solved by enough compute & data.

A lot of the promises of AI don’t sit right with me.

Maybe I’m a hippy but I think a lot of the world is more unknowable than we think, most data isn’t accessible to AI, most things are chaotic.

I just don’t believe, for the foreseeable future, AI will be able to do anything like as many complex things as we think. A lot of limits to the world are the wrapping around it, privacy concerns, regulations, business models.

2) How fast and how does the world really change ?

Technology changes the world based mainly how easily it fits into the world.

In order to fit in, it tends to

1. Find small little cracks to dominate “voice to text”, make an image, write an email

2. Do something thats been done before but cheaper / faster

These two things make it quite unremarkable.

The business case for AI adoption is not particularly strong with tasks.

Well done, the office barbecue flyer went from Clipart to Photo realistic and now it’s a Michael Bay movie.

Change to the world involves taking on Integration issues, existing contracts, secure, technical debt, meetings, fear of losing power, explainability, audibility.

The vast majority of vast software, think of SAP, Workday ,VMWare, Salesforce, Jira, Teams, PeopleSoft, WebEx, Citrix, Sharepoint is absolutely fucking awful. And it doesn’t change because they get flown to F1, and a software license audit will be called the moment they try to take it out.

We are more likely to 3D print our own furniture than develop our SAAS

We’ve spent billions and decades keeping bots off the internet and out of our sites, but now we’re going to let agents with read write permission roam free, get out and stroke a dog.

I’ve got this weird idea that the big division in the world is no longer rich and poor.

It’s people who live online and people who live in the real world.

It’s really, really easy to think that AI will change the world fast, until you place one foot in reality.

AI is a bundle of very different tools that are all quite profound in nature.
In the same way that the internet changed things slowly, AI is a mere inflection point in this journey.

The progress will happen (1) Cheaper/Faster, (2) Better, (3) Different

And the leapfrogs will happen when people rethink workflows and systems rather than applying AI to what they do now.

AI will ultimately be 50% culture change, 30% company design , 20% new workflows, and 10% new tools

There will be many missteps, initially AI will absolutely turbo charge bureaucracy, not eliminate it.

In fact, I feel like the prevailing emotion in the next five years will be enormous tension.

The tension between what we know AI makes possible & what we routinely experience.

We will be told that image recognition can predict crime before it happens, and detect Cancer at scale, while the bag scanners at airports are slower than ever.

We will be told AI and Big data will offer personalized ads, but my Credit card will decline a $10 charge in Germany, a week after buying a $10,000 flight TO Germany

We will lose another Boeing 777 , we will get an MRI scan on a DVD, we will need to do a Wet signature after a phone has scanned 30,000 data points on my face, picking up a rental car will be as awful as ever, all while we’re also experiencing the most magical moments we can ever imagine.

Little Ones.

1. Turbocharging the wrong things?

This piece in HBR is fantastic and very timely

It shows that AI doesn’t so much reduce work as make it more chaotic and intense.

One of the really weird things about the modern era is that productivity still seems related to effort and outputs more than outcomes. The big fear from AI for me, is that we do vast amounts of things that don’t need to happen.

Because focusing on what matters takes balls.

2. What could go wrong with a bot controlling everything?

You’d be an idiot to try Claude bot unless you know precisely what you’re doing. I don’t, so I relied on pieces like this, to show me what I’m missing

3. Personal AI
   
   This is Googles first foray, but I’m sure the next phase of AI will be about trading privacy You share more, you get more help in return. Then someone uses this for ads that you find quite helpful.

4. Book burning at scale.

It’s always a bit weird to me that if you copy one book, you get in trouble, if you scan a million books, burn them, then charge people to make copies of them, you’re fine

How anthropic threw away millions of books

5. Look mum, no hands.

Wonderful human and great friend of mine, Alex Roy did something amazing, well, the computers did.

I’m just putting an AI generated video of my dog in here, because I feel like it. It’s based on a photo, but the video somehow looks nothing like him.

Bites

• Wellness but on steroids, well, peptides Longevity and body hacking is changing

• Audi Naming mess - I don’t get how car companies make so many dumb big decisions.

• OpenAi is quite screwed - One day Sam Altman and Sam Bankman-Fried will have more in common than just a first name.

• For me AI is mostly recursive, It’s based on the (A)verage of the (i)nternet. Can AI create new? And can it do new math?

• The number of scams Meta happily take money from is incredible.

• What It Actually Takes to Build a Keynote” - superb piece from Zoe Scaman

• 3D printed wood furniture

• The paper passport is dying ( slowly)

• 250,000 drone deliveries from a company that should be celebrated a lot more

• Probably the best, and therefore most depressing, fake music.

• Financial nihilism will probably be a big thing this year.

• Learn from the Startup graveyard

• Flying taxis are coming to Florida.

Blast from the past

Some trends I wrote 6 years ago. The future is not that unknowable but the timing is impossible.

Me Me Me

This year is a bit quiet, it always starts off that way.

I’m on the edge of lots of proposals being accepted or declined.

I’ve 15 events that may say yes or no.

So as of now I’m open to:

• Keynote speaking or more intimate powerpoint based affairs.

• Workshops / Training on all elements of AI

• Consulting where I actually solve problems you have or show you new possibilities

• Consultant Agitator- Got McKinsey’s B-team in? Your boss thought Bain could help, but it’s basic? Average thinking from Accenture. Bring me in and spend 5% more to get 50% more out.
  Make them work harder with me as an ally and therapist.

This Newsletter goes out irregularly, but it’s been quite a long gap even for me. I’ve got a 4 month old kid and he’s not really sleeping. It’s not the time it takes that kills productivity, it’s the struggle for clarity.

Deeper ones.

A) Reality Cheques

There are times it feels like although Wall Street isn’t Main Street, and the US isn’t the World, that the entire Global Economy is propped up by confidence from the booming NASDAQ…… which has come about from 7 huge companies sending fake money between each other, and I wonder about where that leads us………

I personally don’t think it’s a bubble, I think it’s a spiral, it’s a circle of promises, all based on the assumption that if you build it they will come……And if they come, then value will be created……And if value is created, it carries on.

But is this all just imaginary money, the classic Vendor financing bubble, is this Cisco in 1999 or Lucent and Nortel or Railway mania, or the electricity boom of the 1900.

The widely circulated estimates are AI, as a whole, would need to generate $500B to $1T annual revenue within the next ten years or so to make all the infrastructure investments work. And thats quite some way away for something mostly used as a search wrapper for Reddit.

The way I think of this cycle is something I adapted from Goldman Sachs.

There are 4 phases of AI, or layers:

1. The first phase is the foundation, the construction of core AI infrastructure. The GPUs ( Nvidia), the Datacenters, the electrical power supply, a friend of mine works for Rolls Royce, a company absolutely swamped by demand for backup cooling pumps. These companies are all printing real money.

2. The second phase, the people building ON this , and the people paying for this, are the Hyperscalers, these are Amazon, Meta, Google, Microsoft, Oracle, Apple, and more, these are the people who think their only way to win or compete in the future is to be the people who own or run AI. In many cases I think they are utterly wrong, Meta is a media company, Apple doesn’t have to have its own AI, but thats just my view.

3. Phase 3 are companies who unlock AI-Enabled Revenue, these are companies who don’t need to build their own AI infrastructure, but can ( in theory) make more money from building products or sell services using think Palantir, ServiceNow, Accenture, Fortinet, Adobe, Workday, MongoDB etc etc

4. And then on top of these Products and services is Phase 4, AI-Enabled Productivity Gains.

Which, in theory, should be pretty much everyone and everything.

The promise has always been that the hyperscalers pay for the infrastructure, The AI-enabled revenue companies pay for the Hyperscalers, and the AI Enabled Productivity gainers pay for AI Enabled revenue.

The only problem seems to be there’s remarkably scant proof that AI Enabled Productivity is that great, and Phase 3 companies are often pushing AI products onto unsuspecting consumers, rather than selling them for huge demand driven sums.

How this all ends or where this all goes is completely unknown. My sense is that the hype has got out of control. It’s like Pascal’s wager, belief pulls people in, because people would rather the same as each other but wrong, be seen to knowledgeable and optimistic and enthusiastic, rather than measured or cautious. We have this bizzare idea that companies who move fast and first (with technology) win, when time and time and time and time again, this just isn’t true.

It’s not that AI or Generative AI isn’t helpful, or magical, or a paradigm shift in possibility, it’s that my beliefs are.

1. It takes depth of application of technology for productivity gains to appear. ( Transformation is about the lever arm, not the force)

2. Which means it takes time ( a decade?) for changes to appear.

3. A lot of really boring work needs to happen before this is possible, like data strategy, organizational change, auditability, explainability, workflow analysis, etc.

B) OpenAI- Everything Everywhere All At Once.

The pace of launches everywhere in AI land is crazy, but OpenAI’s launch tempo is quite absurd.

A new browser used to take a large tech company 3-5 years to launch, agonizing over every detail, plotting out a launch strategy, bug testing forever, but Sam Altman appeared totally perplexed in the quite cute Atlas Browser launch video, like he was seeing the work of people he’d never seen before.

Whether of not there is a public appetite to share passwords with an AI is one thing, whether sites that have spent decades and a fortune keeping out bots, will suddenly do a 180 and welcome them, is far from the certain. AI browsers seem to ask the question, wouldn’t it be great if we could talk to websites and I’m just not sure the answer is yes.

In the last month OpenAI has also

- launched Sora2, an impressive next gen generative video app

-And seems likely to soon to launch a social video app to showcase their own slop.

-It’s launched “buy in ChatGPT” million Shopify merchants, like Glossier, SKIMS, Spanx and Vuori

-It’s launched a quasi App Store/SDK. A new world of Conversational apps you can chat with (and the tools for developers to build them)

-It’s launched a way for companies to use their own knowledge better ( I like this one)

-It’s getting into Erotica

-And the sun that all tech grows towards, Advertising, is a key focus.

-It’s recruiting investment Bankers to help refine its ability to write strategy decks.

And this is far from everything.

Which begs the question, what exactly is it trying to be?

OpenAI as Operating system? OpenAI as a commerce front end? OpenAI as social media network? OpenAI as a media channel? OpenAI as a productivity tool?

It seems quite likely it’s trying to be everything.

Because true to the heart of every tech company is the idea that boundaries between company types mean nothing.

Your core competence is just building something, not category expertise.

Seems quite likely there will be four goals.

1. Make a product thats always best in class and top of mind.

2. Monetize through Advertising ( It has the best intent data you could ever ask for.)

3. Monetize through Affiliate marketing- take a cut of every transaction

4. Monetize through data selling ( It has a rich picture of how people behave and interests)

I’ve got a lot more thoughts on this, but I can’t express them clearly because I’m tired. Join me next time.

But I wouldn’t be so quick to assume that hundreds of millions of dollars will flow through ChatGPT to buy stuff on Walmart, I think conversational commerce is a poor UI.

And while Advertising will be huge for OpenAI, it also risks destroying trust, one of the key attributes that currently makes it sometimes useful.

Shhh I’m talking

I’ve always felt that innovation didn’t really happen in San Francisco. That’s just where the money was. I’ve always felt it happened in places like Bucharest or Tel Aviv or rural India or Brazil.

So I’m thrilled to be speaking at Brand Academy Day. I’m kicking it off on November 13th in Tel Aviv.

Will be talking about AI and Advertising (and other changes) and what it all means, and what to do about it now and soon. Read more here

Little Ones.

• “Screens are clumsy, a necessary evil, an intermediary step. They were never going to last forever”, these pieces really piss me off because the picture says a thousand words. There’s never been a better way to buy things or do things than a visual interface.
  
  Yes, some parts of the internet will become an ambient assistive thin glanceable layer, as I’ve often said, but AI builds on the existing platforms of today.

  
  If we want to reduce screentime, we don’t need a new operating system. We need personal discipline and different incentives that fund the internet.

• Somehow Meta Is obsessed with ramming AI down everyone’s throats, and now AI is starting to dick around with ads and make them massively off brand. I find this thread amazing. People don’t get that the cost of a bad ad isn’t poor metrics, but complete brand destruction.

• There was always a fascinating question about who should make the helicopter like drones of the future, Would it be helicopter makers or Chinese toy companies? Is it easier for a massive company that dominates aerospace to win over faster, more nimble companies with no concept of safety? Or is it better to be neither , it looks like Joby Aviation may be winning.

• Someone made a “TV Channel” of entirely AI generated videos. Mesmerizing and souless.

• I’m quite intrigued by the idea of reducing the frequency of company earnings, In a world where short-termism makes everything a frenzy, it seems like a good idea.

• Introducing AlterEgo, the first near-telepathic interface, designed to make technology as intuitive as using your inner voice. It’s actually a bit of a scam requiring you to “mouth words silently” .

• “If AI Takes Our Jobs, It’s Because We’ve Made Our Jobs Really Stupid”, I was quite honest when I spoke at ADMA in Sydney.

• I’m fascinated by all of these toys that don’t have screens but offer the promise that they’ll help bring up your kids for you. Technology is never good for us or bad for us, It’s all on how we use it, But it’s always very hard to know what’s an extremely good idea and an extremely bad idea.

• I loved this piece on accidental innovation. Of 190 “major” inventions between 1800 and 1970, about 14 could be described as accidental.

• People who know AI, don’t know the real world #3031 , why AI isn’t replacing Radiologists

• Rising trends in boredom, a study by Nature.

• People are increasingly uncomfortable saying, “I don’t know.” , perhaps behind a paywall, so just pretend you read it.

• One of the more promising, leave your smartphone behind, go see the world, devices out there. This is meadow - Leave your phone at home and make some memories.

• Smarthome 2.0, it’s about time someone made a camera that tells you about your poo. The most amazing thing about this is the $150+ a year subscription cost.

So enough about photographing your own shit, let’s put down the internet for a bit.

Have a great day, everyone.

And apologies for the typos, I’m tired.

Tom

Deeper ones.

1 - Rethinking life.

What happens when tech smashes assumptions?

A few years ago I was on a Greek ferry passing one of the countless dreamy uninhabited islands, and it dawned on me that it’s always been near impossible to live in such a place—but that this could soon change.

Fifteen years ago you’d need to somehow get electricity, cellular or fixed phone service, fresh water, and of course you’d need to build the thing. To get stuff (like yourself) there often, you’d have to fashion some sort of harbor. It would pretty much all be impossible in almost every direction.

Today you could plonk any number of futuristic modular, prefabricated homes from a helicopter onto some piles in seconds.

You’d get insanely fast, cheap internet and comms with Starlink.

Electricity would come cheaply from wind and solar, batteries would smooth out supply shortages, and desalinated water would be simple.

You’d circumvent a harbor by using a drone, and you wouldn’t need to leave so much because school or work could be remote, doctors’ visits could be done via telehealth.

The only thing that would be difficult would be every single part of getting any form of permitting, or a mailing address, or permission to live in the country. Literally every interface with bureaucracy would be as slow, painful, physical, and anachronistic as ever.

We often forget that the future is built on the foundations of the past. And while possibilities may change quickly, and our imagination eagerly works at how technology challenges assumptions, things don’t necessarily become faster. They tend to become more chaotic, more tense, more complex. Entropy increases.

This little daydream on a boat a while ago gets to the heart of many things today. But the one I want to dwell on is the idea that technology has profound implications for changing some of the assumptions that hold us back. My first two books were based on the idea that technology changes how we behave and the rules of business. The tagline was: “What would your business look like if you set it up today?”

I don’t think I ever want to write another book, but if I did, the tagline would be: “What would your life look like if you built it today?” Because many of the core parameters of life are design criteria that we never challenge, despite no longer being especially true.

We retire because we presume our bodies will wear out, but most people reading this will be more able to create value at 80 than they were at almost any earlier time. For most people, learning, socialization, and work are unbundled from location. Most of our life is still shaped by status, privacy, scarcity, proximity, risk avoidance, and a fear of standing out. It’s built around formal education, accumulation, consistency, and stability. But who has asked the question: what does a good life look like in the modern era? Are we automatically designing our lives around the parameters of the past? What if we live a really long time? What if economic growth forever doesn’t happen? Are we optimizing for things that we assume matter, but that perhaps don’t mean as much anymore?

2 - Zuckerbored

I do think Zuck is just bored these days. He just announced plans to spend $50bn on a server farm the size of Manhattan to “do AI things.”

While I do find Zuckerberg and everything he touches to be accidentally misanthropic, what puzzles me most about him is he’s remarkably consistent inability to have a vision or good ideas. Other than “the Portal”, buying instagram and buying WhatsApp, he’s never been shown any ability have or follow good ideas.

Meta is a deeply simple and smart company, it makes money from eyeballs, or more specifically from showing ads to eyeballs, or even more specifically, from using closed loop data signals that showed advertisers their ads worked, and automatically debiting their credit cards.

It’s true that today every company “needs” to be an AI company, just as in 2015 every company “needed” to be “a tech company”. But the real question is: how much of the technology do they actually need to own, invent, or operate?

BMW, Netflix, or P&G didn’t become digital by running the internet; they just had to understand it, and it’s implications, and to use it.

So maybe, at the risk of sounding old-fashioned, Apple, Meta, Amazon, SAP, Tesla, Pinduoduo, or Samsung don’t need to build AI from the ground up. Maybe they just need to get extremely good at understanding what it means, and how to use it best and maybe it won’t change consumer behaviors, maybe we like buying things.

3 - My punchiest podcast ever.

I do about 5 Podcasts a monht and much to every hosts annoyance I almost never promote them because can you imagine how insufferable my posts would be. This is an exception because

1. It’s important to me for people to see it. I care.

2. The questions were great, great hosting Jon.

3. I think I was really lucid and clear and articulate for once.

   ---

   The Spark of Discovery – Reigniting The Emotion of E-commerce - Sponsored by Criteo.

   There have always been two types of commerce: shopping vs buying.

   Shopping is about a fun experience, it’s about discovery, persuasion, seduction, surprise. It’s the trip to the mall, the farmers market, it’s something you look forward to.

   And then there is buying, the efficient procurement of a need, the urgent fulfillment of a practical need.
   

   AI won’t change everything immediately but it looks likely to impact both commerce journeys.

   • For shopping it helps us make decisions, it provides guidance, it can make what was overwhelming more fun.

   • But buying could be revolutionized more readily. In theory more logical purchasing could be done in a more automated way. More data, more decisioning also means retailers and buyers can transact in ever more efficient and speedy ways. 
     

   In Criteo's report, “The Spark of Discovery – Reigniting the Emotion of Ecommerce”, the research explores what it takes for brands to stand out when shopping and buying blur across physical and digital spaces.

   For me, the biggest misunderstanding in online marketing is that people don’t want to be helped, remembered, or served. In reality, they expect brands to understand them, to anticipate needs, suggest relevant options, and remove friction. People don’t hate personalization they hate when it's done poorly. 

   The report shows how we can break out of the loop and bring real discovery back to commerce. The companies that succeed won’t be the ones with more data, but the ones that use it thoughtfully alongside AI. 

   Definitely worth a read for anyone rethinking what online shopping and buying could be. 

   ---

   Little ones.

   a) I like these bite-sized ideas. I keep sneaking them into articles, but nobody seems to pick them up—so I’ll just keep hammering on about them.
   

   1. Work expands to fit the technology available.
   We always think new tech will mean shorter working weeks or radical productivity gains. In reality, it usually just enables more pointless work, more bureaucracy, and more distraction.
   

   2. We apply technology where it’s easiest, not where it matters most.
   The first use of any tech is to do the same things faster or cheaper—it has to slot into existing systems, mindsets, and business cases. The next stage is doing the same things better. Only much later do we rethink the system entirely and use it to do new things in new ways.

   
   3. Most predictions are just preferences.
   People imagine the future in ways that suit them, shaped by their worldview and biases.

   
   4. AI finally forces ambition.
   For once, companies see their IT departments as strategic, future-focused functions—not just cost centers.

   
   5. Tech adds entropy.
   It rarely replaces old ways of doing things; it just layers new ones on top. The touchscreen didn’t kill the keyboard. Our wallets have more in them now than ever.
   

   b) A nostalgic history of pizza ordering online , somehow Pizzas and Porn seem to be the harbinger of tech.

   c) It’s great when anyone can make anything, but sometimes maybe they shouldn’t What are people actually “vibe coding” and have a look and see if it’s any good.

   d) Wouldn’t it be great if Amazon could hear everything I ever said

   e) How to rewire your curiosity, just one of 129 tabs open on my laptop, one of which is about oddly short runways.
   
   f) A site full of amazingly useful newsy charts.

   g) I got more drunk than planned with a Norwegian ( Hi Vidar) on a drink that crosses the equator twice and it fascinated me how inefficient logistics can be a powerful product proposition
   

   h) Worst advice ever, the UK government genuinely and seriously tells people to delete emails to conserve water. ( Aren’t most data centers closed loop anyway?)
   

   i) Host makes an Airbnb damage claim using AI generated images, what a funny time to be alive.
   

   j) Cartwheels and somersaults are easy but Robot dexterity is still hard.

Me me me.

I’m doing more “board level advisory” work these days, which is a way of saying helping boards or individuals at the top level, navigate change, challenge assumptions, avoid distractions, bring the world into the room, and ask more naive but powerful questions.

I’ve got a couple of open slots now, so please get in touch if I could be valuable.

That’s it for now.

Thanks,

Tom

( as always, I read all replies, and reply to most )

Wow, it’s exhausting out there. Every time a new technology emerges, I get genuinely excited about what’s possible. But then the noise kicks in. These days, it pays to be loud and wrong, bold and clueless, to stir panic rather than understanding. Futurists get paid to be interesting more than helpful, so I’m not one of them.
Here’s my latest attempt to bring a little nuance back and be useful.

Deeper ones.

1 - Robots will take your chores.

I don’t know that much about (humanoid) robots but I do think they represent a really interesting tension in today’s world.

These days I see ever more videos of ever more amazing machines, with the promise of performance increases, price decreases and self-learning technology which makes them easy to use, to improve and to adapt and program.

In no particular order, some thoughts.

1. Something about these videos always looks really fake, maybe it’s skepticism born from distrust of a “fake it until you make it” era, or maybe the filming technique is weird.
   

2. They always looks much more like people than makes sense. If you wanted to make a really good robot, you’d start with wheels and low to the ground, making them human seems to freak people out, but it does have the massive bonus of making them MUCH easier to fit into a world that we’ve primarily built around people ( and cars). This is how technology starts, it fits into physical and mental spaces we have already.
   

3. It's quite easy to get excited, just imagine a home robot that does everything you ask for $20,000. No more ironing, cleaning, trips to get milk, no more changing the bed. Just imagine what this can do for people who need carers, the elderly, just imagine busy people with kids working 3 jobs, now able to rest a little. Imagine what this can do for productivity. One of the big questions of the current AI era is “what are we going to do with all this free time”, a question that always seems hilarious in retrospect.

4. At the same time, like most technology of this era, it makes me wonder if we’re being led by techno-optimizers, the sort of people who think food should be a tablet, friendship is more efficient if done online, that alcohol is illogical, the sort of people who want to live long, narrow, optimized lives, devoid of the inefficiencies that many call “living”.
   
   Increasingly I tend to think that very little of life is something to outsource. It’s quite nice to fix the garbage disposal unit, it gives me a sense of mini progress to make the bed, cleaning is my meditation, cooking is my therapy.
   

   

   
   

2- The Ambient Internet.

For the first time in 8 years, I went on Facebook the other day. I got absolutely addicted to Facebook Marketplace, but also mindlessly scrolling. Here I was with a new baby, new puppy, a great wife, and I was mesmerized by videos of snooker, boats going through choppy water, courses on how to make $50k a month using AI, literal car crashes, soccer technique, and AI-generated women with fake fake breasts.

Technology people have decided the phone is the issue. It would be better if we wore the Internet on our face, or if cameras were on glasses, or that something could overhear everything we said and interrupted us. It's almost like no one remembered the complete failure of the Humane Pin.

Alas, here we are again, a new “smartphone killer”. I wish people would realize that the problem isn't the form factor of the phone, the problem is the incentives that drive software and services or at least the stupid KPIs.

The only thing Netflix should really care about is how happy people are with the service. Instead, they take time spent on Netflix as an easy-to-measure proxy for happiness, and then do everything they can to increase time spent, by automatically suggesting incredibly bad shows. But because they autostart, we become transfixed by the mundane cheap crap that algos know we can’t stop watching.

If we want to fix the screen addiction problem, we need to rewire the industry, not the device. And we also need some discipline.

3- The Agentic Era is probably nonsense

The world is really complex, with old software, technical debt, data is all over the place, databases that can't talk to each other, and complex security, permissioning, and governance.

When Web3 came along, it was driven by the hope that new plumbing would change all of this, but it turned out life was more complicated than that. Now we have AI agents; the dream is that they provide a new interface onto the old web.

Rather than my using Condé Nast Traveller to fall in love with Montenegro, Kayak to find a flight, and BA.com to buy a ticket, I’ll just say to a bot, “Take me somewhere nice,” and it will handle everything.

A couple of things hit me about this.

• Most people quite like shopping for most things.

• The rationale that people use to choose things is often quite unexplainable; Zurich or Madrid airport makes me happier for transfers than Heathrow.

• The BA website is almost impossible to use and always down. Why would a robot be luckier than me?

• Most companies have spent years keeping bots off (CAPTCHAs, anyone?). Why would they now let the most powerful bots ever have free reign

• Every demo of an agentic flow didn’t really work.

• Everything online is always changing; the moment anyone in the chain changes anything, the system could fail.

So I was very unsurprised to hear that Cloudflare is blocking bot access this week

Podcast of the month

Cybersecurity used to be a technical backroom topic, but in the age of AI, it becomes a topic for everyone in the boardroom.

It’s great to “move fast and break things”, but security risks are existential.

As AI accelerates threat complexity and the attack surface grows by the second, staying ahead isn’t optional.

That’s why I’ve been tuning into Threat Vector, a podcast from Palo Alto Networks that goes beyond the headlines, and into depth with a rare thing, deep experts with clear advice.

It’s sharp, fast, insightful with real perspectives from industry leaders tackling real problems.

If you care about the future of security, trust, AI, or the future survival of your company, this is well worth your time.

Check it out at https://www.paloaltonetworks.com/podcasts/threat-vector

Shallower Ones

1 - Magical to Mundane in seconds- I’ve been slow to try Waymo. It happened this week in San Francisco; it was both the most magical (the first 90 seconds) and the most boring (the next 13 minutes) experience I’ve ever had.

It felt quite lonely and sad, and it was precisely the same as any Uber I’ve ever had, just without the feeling someone who deserved it was getting paid. As with many technologies, the initial impact isn’t huge; it just fills the same gap. The real change will come when robotaxis thrive in rural areas where Uber makes no sense, when cities turn car parks into thriving spaces, when cars communicate to reduce congestion, and when buses become intelligent and dynamic. It’s the second- and third-order impacts that will be enormous.

2- Gene editing will probably be one of the biggest issues of this generation, we've been saying that for a long time, but you can just see it happening now. Turbocharging and entrenching privilege forever?

3- The Internet made the cost of distributing stuff tend towards zero, AI has made the cost of producing stuff tend towards zero, an algorithm spread content far and wide. Content creation has basically become a casino, where every spin of the reel is free, and people have been making a fortune making endless videos for YouTube, or images for Pinterest, or fake news for Facebook, and nobody seemed to care, until now. I’m so happy YouTube is banning slop

Wider Ones

Why shops make a city great

Economic miracles stem from discovery, not repeating tasks at greater speed

We have no idea how to value companies.

Calm down, why FOMO isn’t a strategy

A raspberry pi, but for robots. Good work Hugging Face

UI Trends Mood Board for 2025

The rise of the Japanese toilet

The world's first Hydrogen economy?

Why is modern commerce corrosive?

Half the money I spend on bureaucracy is wasted

How does United Airlines pick weird places to fly to?

Older Ones

A deck I made about Trends in 2015, I’m mega proud of it.

Me me me

Right now I’m focusing on a new Workshop offering with Chris Johns of Bttr.

It's designed to help companies really understand what AI means for their business, and what to do about it.

I'll be talking more about it in the future, for anyone interested, get in touch.

That’s all for now.

It all feels a bit overwhelming these days. My attempts to keep up to date with the daily developments in AI have been exhausting, which explains the irregular nature of this newsletter.

“Nobody knows anything" was coined by William Goldman, a screenwriter and novelist, as a way to explain how Hollywood “worked”, and it always seems like a perfect description of the current AI world, the more you read, the more you experiment, listen, talk, learn, the more it seems to make less and more sense.

I’ve learned to take a step back and try to see things in context, over a longer period, then some patterns but also questions emerge.

Big Ones

Where are we on the AI hype curve?

Is it the law of diminishing returns or exponential growth, Zeno’s paradox or the runaway effect? Where are we?

It’s easy to get swept up in the wave that is AI. I returned to Facebook to buy rare plants from locals on Marketplace, but my entire feed was flooded with ads for AI courses from prestigious institutions or tips on how to make money as an “AI transformation specialist.” It felt rather scammy—very web3 or metaverse. My Twitter feed is an endless stream of proclamations: vibe coding means anyone can do anything, universities are dead, or AI-first startups will eat the world, despite employing just two people—one of whom might be an enthusiastic 8-year-old.

There’s plenty of evidence to the contrary, too. Klarna rode the PR wave of replacing humans before almost entirely walking it back. Models that once seemed to improve dramatically are no longer doing so—or is this a classic case of our expectations of technology advancing faster than the technology itself? What’s magical becomes ordinary, then disappointing, in about 43 hours.

Three questions in my head:

1. Not so fast, the world is messy.
   There’s a certain Gell-Mann Amnesia effect with AI: the less you know about a field, the more impressive AI seems to be at it. It’s easy to think it can replace a lawyer, architect, or copywriter if you have no idea what these jobs actually involve. The only exception seems to be coding, where many software engineers are the most impressed by what “it” can do. Regardless of its ability to handle specific tasks, we often forget that most jobs aren’t just tasks—they’re far messier. The role of a receptionist, for instance, probably involves 24 different things, of which an iPad can replace only one.

2. We know how the movie ends……or do we?
   If there were one chart I could make go viral, it’s this: technology has always progressed in three stages—(1) cheaper and/or faster, (2) better, and then (3) different.

   
   First, it’s used to make the same things, but cheaper: the plastic button, the steam engine, the computer—all were initially used to do the same things, just more cheaply.
   
   Then, it’s used to do the same things in a slightly different but better way: the electric lightbulb, the electric iron, synthetic dyes, steel.
   
   Finally, it’s used to create something radically different: film, microwaves, lasers, the TV set.

So, we kind of know how AI will go:

• 2025: Automate customer service

• 2027: Improve customer service

• 2030: Reimagine customer service
  

• 2025: Automate ads

• 2027: Create much better forms of advertising

• 2030: Redefine what ads should be

  
  etc

  

  3. There are really three types of companies when it comes to AI adoption
     

  Simply sprinkling AI onto a legacy process won’t work as well as redesigning that process from the ground up with AI at its core.
  
  The differences between these three types of companies are stark:

  • A large, traditional multinational trying to adopt AI.

  • A large, digital-first firm aiming to integrate AI.

  • An AI-first startup built to leverage it from day one.
    

  AI can be deeply transformative, but it’s incredibly hard to get big, established organizations to change. Even famously nimble, entrepreneurial companies in the second category will struggle with challenges like security, data governance, and new workflows.

  So, yes, AI-first startups (type 3) will likely harness AI’s power first. But then the question becomes: you’ve mastered AI—do you know how to make money in your sector? Will an AI-first retailer, bank, insurance company, energy firm, or marketplace automatically dominate? Probably not. Most winning companies succeed by understanding their sector deeply, not just by mastering the tech.

Non -AI things.

Make Advertising Great Again.

I got a column in the drum, in my second post I swore about how shit ads are now and how we can Make Advertising Great Again.

”To be advertised at today is to be insulted, the promise of adtech is a lie, we don’t offer one on one conversations at scale, we offer random precise guesses as to who you are, that result in it thinking you’re an old, pregnant male suffering from poor health. It will engage you in conversational Cantonese to stoke your passions for luxury handbags and industrial rivets.

Take a few seconds to unblock those ads, and then take a look and see what you get. Did a human see this? Did anyone care? Did anyone expect me, a human, to click?

I don’t think we realize how deeply nasty it all is; between the lies, the fraud, the scams, the bots, the targeting, the following, the harassment, the automated ways to spank budgets on fake clicks, faster than ever. It all makes pig butchering or 419 scams look rather sophisticated and above board” - Read more here.

Sponsored post ( but still good) - Spinnaker technology

Nokia. Kodak. Blockbuster. Borders.

Remember the relentless tech disruption narrative from 2010?

These companies were cautionary tales, and the question on everyone’s mind was who’s next?

Surely Walmart & Target were doomed, Airbnb would crush Marriott, & Toyota would fall to Tesla, Uber, Waymo, or all three.

Who needed a Gym in the age of Peloton, or Bank when we've got an neoBank.

But then… nothing much happened.

For all the talk of “Disrupt or Die,” “Digital Darwinism,” and “Creative Destruction,” reality turned out far less dramatic.

Carry on reading the post here.
https://www.linkedin.com/posts/tomfgoodwin_nokia-kodak-blockbuster-borders-remember-activity-7287526658012565504-D1fU

Little ones

• An interesting piece on various companies completing changing how things are made.

• Love this piece on Cognitive debt. It’s when you skip deep thinking to get quick answers, but you don’t understand why those answers are correct. It’s like borrowing knowledge without paying the “interest” of truly grasping it

• Robots flopped at a half marathon, proving they need wheels, not legs. Human-like robots are pointless unless deception’s the goal. Why do people keep forgetting this?

• Quite a useful case on how a Legacy Financial Institution Went All In on Gen AI. HBR is often nothing like as good as it used to be, but this is decent.

• A lot of the time it seems like the jobs the US wants to bring back don’t really exist any more and that’s prob a good thing, coal mining anyone, but sometimes it’s not that simple. Turns out robots making shoes isn’t that easy.

• I love how you can art from the weather in real time with this , just wish people made Digital Art Screens for this stuff.

• I’m always fascinated by bulky, large, outdated infrastructure. Sometimes I do think that deep transformation is essential, but there's never a business case to be made for it. Things like Covid showed sometimes urgency, desperation and disruption is a good combo. Doge’s work is quite interesting to show the chaos that ensues. Sometimes you have to Take steps back to go forward

• People seem to think startups like this will threaten management consulting, I think it’s nonsense, people don’t choose McKinsey for ideas or work done, they choose it for career insurance and to show you tried , and to get existing work sold in by outsiders.

Thought for now.

What are we saving all this time to do?
The more technology I have in my life, the most determined I am to be bored.
I’m cooking more than ever, doing more things slowly, I’m getting surprisingly obsessed with house plants and watching them grow. I also have a puppy and watching him attempt attack a tennis ball gives me a rush that no notification can currently compete with. So much of live isn’t to be automated or streamlined, it’s to be entranced by. Miami is about to enter the season of the very best clouds, it’s exiting.

All about me.

I’m speaking at AntiCon 2025 in London this week about AI and Advertising.
Then some private events in Paris before another talk in London for Mailchimp on June 12th.

Currently planning a bit of a global tour to Australia, maybe Cape Town and somewhere else on the way in August 2025, and prob the Middle East in November, if you want to add me en route.

This is my puppy. He owns me now. He’s called Arlo. He’s tiring but magnificent. Everyone should have a dog.