What Doesn’t Change in Marketing in the Age of AI (Part 4)

Email, the Missing Relate Layer, and NeoMails – 2

AMP changes the category

Before introducing NeoMails, there is a technical shift worth naming: AMP for Email.

Most people still think of email as a static object — text, images, links, a call-to-action button leading somewhere else. AMP changes that category. It allows forms that submit inside the inbox, polls that respond in real time, quizzes that react immediately, dynamic content that updates on open, and actions that complete without the customer leaving the message.

A static email is a document. An interactive email is a product.

That shift is more significant than it first appears. A surface that can receive input and respond to it is a fundamentally different kind of thing from a document that can only be read. The implications for what email can do in a customer’s daily life are significant and mostly unexplored. NeoMails are built on this infrastructure.

NeoMails as the attention architecture

NeoMails are not a better email template. They are a different category of inbox interaction — one designed for the conditions AI creates.

The unit is the APU: BrandBlock, Magnet, Mu, ActionAd.

The BrandBlock gives the NeoMail identity. The brand’s voice, perspective, and world — before anything is asked of the reader. The Magnet gives it lift: a quiz, a prediction, a preference fork, a tiny challenge that earns engagement before anything is asked of the reader. Engaging with the Magnet earns Mu — the micro-reward currency that builds a visible streak and accumulates over time. The ActionAd funds the send — one curated, brand-approved in-email action unit per NeoMail, covering the cost of delivery and making the Relate message economically rational for the first time.

This architecture is exactly suited to what AI changes and what it does not.

AI makes content abundant. NeoMails are not mainly a content idea. AI makes personalisation cheaper. NeoMails are not mainly a targeting idea. AI makes creative production easier. NeoMails are not mainly a creative-format idea. They are an attention architecture — bounded, participatory, cumulative, self-funding.

Bounded: sixty seconds, completable, with a beginning and an end. There is a cognitive science finding worth naming here: the Zeigarnik effect. Incomplete tasks occupy working memory; complete tasks release it. An infinite scroll produces a form of low-grade anxiety — the sense of having consumed without finishing. A completed NeoMail produces something different: the mild satisfaction of having done a small thing. That satisfaction is not incidental to habit formation. It is the mechanism of it.

Participatory: the Magnet requires the human to do something — predict, choose, rate, answer. Participation generates a first-party signal that no AI-generated content can replicate. A choice expressed, a prediction staked, a preference registered: these are human signals of genuine attention. In an AI era, where passive exposure becomes cheap and abundant, active participation becomes the premium signal.

Cumulative: Mu builds a visible record of the relationship. From the customer’s point of view, the inbox usually has no memory — each send is a stranger introducing itself. Mu changes this. The balance in the subject line says: yesterday mattered. Showing up left a trace. The relationship has a history. Histories are harder to abandon than novelties.

Self-funding: the ActionAd covers the cost of delivery. A message that earns attention is also the message that funds itself. That single economic inversion is what makes the entire architecture viable at scale.

The Relate message as the human signal

When AI generates all the transactional and promotional content, the brand that sends something genuinely worth opening for its own sake stands out precisely because it is rare. A message that asks for nothing except a minute of genuine engagement becomes the rare signal that a human relationship is being maintained — a brand that is treating the customer as a person rather than a conversion target.

That is not a sentimental argument. It is a competitive one.

In an inbox increasingly populated by well-crafted machines talking at customers, the Relate message is the differentiator. Not better AI. Not smarter targeting. Simply: something worth opening when nothing urgent is happening.

Email did not fail because the channel declined. It failed because brands stopped having anything worth saying between transactions. NeoMails are the answer to that failure. And they are more relevant in an AI world than in the one that preceded it.

Thinks 1954

Manu Joseph: “Spare me the young, why should they matter for everything? They have no money, no clout. Even if it was true that they won’t read a magazine, which I don’t believe is true, why should that decide the survival of a product? I notice this fixation with the young in several businesses. It is as though there aren’t other kinds of people on this planet. That is odd when those businesses survive on other kinds of people. In cinema, television, dining, apparel, hospitality, social media and just about any business except hospitals and old-age homes, people at the helm worry a bit too much about the young—how to get them, or how to keep them. Entire nations are obsessed with the young. That insufferable buzzword, ‘demographic dividend,’ is all about this phenomenon. The value of the young is not only their fertility rate anymore, as the average age at which people become parents has been increasing. Their value is said to lie in their contribution to society, which I believe is overblown. They are merely loveable, and society is coming up with excuses to disguise its love for an adorable segment as a wise investment of time and money.”

Benoit Denizet-Lewis: “I’ve come to believe that the “self” in self-transformation is only half the story. Change is less about willpower than we imagine, more shaped by other people than we admit, and far more mysterious than the self-improvement industry can afford to sit with…In other words, we might think of self-transformation as a team sport.”

TheMaxSource: “In the last 12 months, something shifted in the tech-related job market and it didn’t make headlines. Just a quiet, consistent pattern showing up in job listings across tech related industries: ”The specialist titles are shrinking, and the broader ones are coming back.” Not because companies got lazy with job descriptions. Because the economic logic that created hyper-specialization in the first place — that scale required narrow focus — no longer holds the way it did. AI absorbed the repetitive execution. And suddenly, one person with range is worth more than two people who each own a slice.”

WSJ: “The idea behind silicon sampling is simple and tantalizing. Because large language models can generate responses that emulate human answers, polling companies see an opportunity to use A.I. agents to simulate survey responses at a small fraction of the cost and time required for traditional polling. Phone polling has become exponentially harder. Web polling is too uncertain. Silicon sampling removes the messy, costly part of asking people what they think. But this undermines the very idea of the opinion poll. Public opinion is used to guide policy, politics and social science, and it has value only insofar as it summarizes the beliefs and opinions of actual humans. Using simulations of human opinions in place of the real thing will only worsen our broken information ecosystem, and sow distrust. We should not turn to an artificial society to try to understand our real one.”

What Doesn’t Change in Marketing in the Age of AI (Part 3)

Email, the Missing Relate Layer, and NeoMails – 1

Why email survives every wave

Email has been declared dead many times. It was supposed to be replaced by social networks, then messaging apps, then push notifications, then mobile apps. Now, perhaps, by AI interfaces and agents.

And yet it remains.

That persistence is not nostalgia. It is structural. Email survives because it combines a set of properties that no other channel has managed to replicate together.

It is identity-linked: tied to a real, named person rather than a device ID, a cookie, or a probabilistic inference. It is portable: the same address works across every device, platform, and application, independent of any single ecosystem’s fate. It is permissioned: the customer chose to share it. It is universal: it requires no particular app to receive. And, crucially, it is not primarily governed by an external feed algorithm.

A social follower can disappear behind ranking logic. A mobile app can be deleted. A cookie decays. A device ID breaks. An email address persists. In a world where rented surfaces grow noisier and less reliable, those properties become increasingly valuable rather than increasingly quaint.

The problem with email is not that the channel failed. The problem is that the product built on the channel became too narrow.

The product failure: Sell and Notify only

Most brands use email in only two ways.

They Sell — the campaign message: launch, offer, discount, cart reminder, seasonal push, urgency, conversion. And they Notify — the transactional message: order confirmation, shipping update, password reset, account alert.

Both matter. Neither is enough.

Because most of a brand-customer relationship does not happen at the moment of purchase or immediately after it. Most of it exists in the long middle: when the customer is not actively shopping, not waiting for a receipt, not browsing the category, not in-market at all.

That is where the relationship either stays warm or starts to cool. And conventional email has almost nothing to say in that period except more extraction.

If every email either asks for something or confirms something, the inbox becomes a sequence of interruptions organised entirely around the brand’s agenda. The customer learns the pattern. They stop opening unless the offer is strong enough or the transaction is important enough. The relationship loses rhythm. Attention decays. Eventually the same customer appears in a Google or Meta auction — the brand pays twice and wonders why retention feels so weak despite all the automation.

This is not a channel failure. It is a product failure. The channel was reduced to two modes when it needed three.

The missing layer: Relate

What is missing is a third message class. Not Sell. Not Notify. But Relate.

A Relate message exists to keep the relationship alive between transactions. It does not need a campaign to justify it. It does not need a purchase to trigger it. It simply needs to be worth opening for its own sake — because it gives something before it asks for anything.

That sounds almost modest. It is in fact the structural missing piece in modern brand communication.

The absence of Relate explains more of marketing’s persistent problems than any other single factor. It explains silent drift — because customers have no reason to stay engaged when they are not buying. It explains the reacquisition trap — because brands that never Relate must reacquire attention through paid channels when they need it. It explains the episodic traffic problem — because a brand that only sends Sell and Notify messages only generates sessions during campaigns.

AI makes this worse, not better. If AI generates all the promotional and transactional content — and it will, reliably and cheaply — the brand’s communications become indistinguishable from a machine talking at the customer. More personalised, perhaps. More timely. But still fundamentally asking or confirming, never simply present. The Relate message becomes the rare signal that a human relationship is being maintained. In an AI world, that signal is differentiating in a way it has never previously been.

Thinks 1953

FT: “[Francis] Bacon articulated a new attitude to nature. As he famously wrote: “Knowledge itself is power.” Nature was not to be revered but interrogated, understood and ultimately controlled. His focus was what he called the “relief of man’s estate”: the systematic enlargement of human knowledge and human prosperity. That foundational ambition to “master nature” is arguably one of the most consequential ideas in history, for good and ill. It underlies the agricultural and industrial revolutions. It also sits behind modern demographic and ecological crises — and the technological revolution of our own age. Yet mastery, in Bacon’s sense, always invoked ambiguity. To understand nature is to gain power over it — but also to become newly dependent on the systems we create. Modern societies are not only masters of nature; they are also entangled in vast technological networks they can neither fully predict nor easily control.”

NYTimes: “Body weight workouts are a convenient and inexpensive way get in shape. But it’s easy to get bored with them or start to plateau. If you want to train your balance or build more explosive power, there is a simple way to level up your exercises by using a standard household step or even the curb outside. “That little bit of height can make an exercise easier or more difficult,” said Dr. Kyle Lau, team physician for the athletics department at the University of California, Los Angeles. For example, you can place your hands on the step to make push-ups easier. Or you can elevate your feet on one or even two steps to make them more challenging.”

WSJ: ““The diary’s dailiness sets it apart in the self-writing sea,” Ms. Rubiner observes. “While a memoir or formal autobiography aims to offer a retrospective story, the diary typically doesn’t, or can’t, because it is written from the middle of an unfolding life.” Having said as much, she hastens to clarify that entries needn’t be made daily for a diary to qualify; nor need the diarist use a physical notebook (some use video or apps); nor indeed need a diary contain personal reflections. The diaries of some consist of little more than, say, a dispassionate daily notation of the weather (my grandfather). The diaries of others (Virginia Woolf, Sylvia Plath) are rich, intimate and self-searching. Still others have charted terrible and turbulent periods of history, the most famous of these being the World War II writings of Anne Frank.”

Neil Borate: “Almost a decade after Coffee Can Investing, Saurabh Mukherjea says the idea is dead. Consumption has crawled. Those fabled moated compounders aren’t compounding. And India, he warns, is heading for a 1991-style crisis.”

What Doesn’t Change in Marketing in the Age of AI (Part 2)

Attention, Push, and the Permanent Asymmetry

The asymmetry that predates the internet

The most important fact in marketing is older than the internet.

The brand needs to reach the customer. The customer does not need to reach the brand.

Everything else is built on top of that.

It is a simple asymmetry, but it explains almost the entire history of marketing. Print, radio, television, direct mail, search, display, social media, push notifications, email — each new system is a new attempt to solve the same old problem: how does a brand get into the customer’s field of attention when the customer was not already looking for it?

That asymmetry does not disappear because AI arrives. It intensifies.

What AI does to scarcity

Content generation becomes easier, which means content supply rises sharply. When supply rises toward infinity, scarcity moves somewhere else. It moves to the human side of the equation. Trusted human attention becomes more valuable precisely because generated content becomes cheap.

This is the paradox of the AI era. Most people think AI will solve marketing’s hardest problems because it can generate more content, more quickly, more cheaply, more personally. In reality, AI solves many production problems while leaving the central relationship problem untouched. It may even make that problem worse. When every brand can generate competent, personalised content at scale, content stops being differentiating. The rare thing is no longer production. It is attention willingly given.

A person can only actively notice, remember, and respond to a limited number of brands and relationships in a given period. App usage data is instructive: people use roughly nine apps per day regardless of how many are installed. Inbox behaviour shows the same pattern — people read far fewer newsletters than they subscribe to. The human attention budget has always been finite. AI does not expand it. It merely increases the number of entities competing for it, while making each competitor more capable.

Why push doesn’t disappear

There is a fashionable tendency to assume that AI agents will make push obsolete. The story goes like this: customers will ask their agents what to buy, agents will compare options, and brands will compete inside agentic marketplaces rather than through direct messaging. Pull wins; push fades.

There is partial truth in this. Agents may become excellent at pull — at responding to expressed intent, researching options, and completing transactions on the human’s behalf. But brands do not live only in moments of pull. They also live in the vast stretches of time when the customer is not actively searching, not in-market, not comparing, not deciding. Pull handles active intent. It does not solve the problem of staying present when intent is absent.

That is what push is for. Push is how a brand reminds, signals, maintains, and accumulates familiarity between purchase moments.

And here is the more important point: even when agents mediate downstream decisions, those decisions are shaped by upstream human preferences, associations, and familiarity. An agent instructed to find the best coffee subscription will not choose randomly from an undifferentiated set. It will act on the preferences and brand familiarity its human principal has built up over time. Attention is upstream of the agent’s instruction. The brand that has maintained human attention wins the agent’s recommendation. The brand that has been absent does not.

AI moves the conversion downstream. It leaves the attention problem exactly where it is.

The owned channel premium

As AI floods rented surfaces — search, social, display, feeds, programmatic placements — with more content and more competition, the value of a direct, permissioned, algorithm-light channel rises. A channel where the brand does not need to win a bid every time. A channel where the message does not depend on an opaque ranking system. A channel where identity is known, not inferred.

This is not sentimentality about old media. It is arithmetic. When rented reach becomes more competitive and more expensive, owned reach becomes relatively more valuable. The brand that controls a direct line to a customer — a line the customer chose to open and can choose to close — holds something that no amount of AI-generated creative can substitute for.

What compounds

Here is the question most of the AI conversation is not asking: what compounds?

Production does not compound. A relationship does.

A customer who has encountered a brand consistently, positively, and meaningfully over six months is in a fundamentally different state from one who saw a perfectly generated ad yesterday. AI can create the ad. AI cannot manufacture six months of accumulated familiarity in one moment.

That difference matters enormously in marketing because brands are not built only through conversion events. They are built through repeated contact, remembered presence, low-pressure continuity, and trust that grows invisibly before it becomes measurable.

Attention compounds. Owned channels compound. Trust compounds. Memory compounds. AI can help produce the pieces that travel through those systems. It does not replace the systems themselves.

In an AI-saturated world, the signal that a human is genuinely attending to a brand becomes more valuable, not less. The company that can create, keep, and deepen that attention without paying a platform tax each time is building on something far more durable than content abundance.

So the question is not whether to use AI. Of course brands will use it. The question is whether, while everyone else uses AI to produce more, you are also building the layer that compounds.

That layer sits where attention, push, and relationship continuity meet. Which owned channel is best suited to host it?

That is where email enters.

Thinks 1952

Andy Kessler: “If you haven’t figured it out, the main export of the U.S. is our standard of living. It isn’t in decline but the envy of the world, hence the rush to our borders. You won’t find that export in economic statistics, but it drives demand for our technology, medical practices and more. The other thing we export is freedom, which drives innovation and lifts living standards elsewhere. Global growth and productivity will be so strong that we’re rapidly inventing robots and artificial intelligence to handle logistics.”

Sunder Pichai: “I’ve always internalized speed. Let’s call it latency for this purpose, and as one of the distinguishing features of a great product. Also, it almost always reflects the technical underpinnings of the product having been done well. There’s a different speed which matters, too, which is the speed of shipping and iteration and release cycles. Both are important.

WSJ: “Forget the keyboard and the mouse. In 25 years, most people will be using brain-computer interfaces to control devices with their thoughts, says Bin He, professor of biomedical engineering at Carnegie Mellon University. These interfaces will be able to interpret brain activity and convert people’s intentions into commands that a computer can understand. “The brain-computer interface will become a technology like the smartphone, where the vast majority of people have one,” says He. “It will make everything so convenient: You just have a thought, and then you control your environment.” In 25 years, He says, billions of people will be using brain-computer interfaces to do everything from messaging friends to switching on the lights and making coffee.”

Greg Brockman (OpenAI): “When we look at the list, there’s consumer, which you can think of it as many things, but there’s a personal assistant — something that knows you, that’s aligned with your goals, it’s going to help you achieve whatever it is that you want in your life. There’s also creative expression and entertainment and many other applications. On the business side, maybe if you zoom out, it looks more like: You have a hard task, can AI go do it? Does it have all the context to do all these things? For us, it’s very clear that the stack rank includes two things at the top. One is the personal assistant, the other is the AI that can go and solve hard problems for you. And when we look at the compute we have, we are not even going to have enough compute to fund those two things. And then once we start adding in many other applications, many other things that AI is going to be very useful for and is going to help people with, we just can’t possibly get to all of them.”

What Doesn’t Change in Marketing in the Age of AI (Part 1)

AI changes the top layers of marketing. NeoMarketing is built on the layers that endure.

Prologue: A Confession, and a Question

A confession before we begin.

I know this series swims against a strong current. Almost everyone in marketing technology today is focused on AI — agents, large language models, generative content, autonomous campaigns. Here I am writing about email, attention, and a sixty-second inbox interaction. I have asked myself more than once whether this is conviction or simply the kind of obsession that mistakes familiarity for insight.

The honest answer is this: I believe the most important opportunities in any technology transition are found not by asking only what changes, but by asking what does not. The crowd is usually right about the direction of change. It is much less reliable about what that change leaves untouched. This series is a bet on what gets left untouched.

That does not mean AI changes nothing. It changes a great deal.

Content creation becomes near-free. A brand that once needed teams of writers, designers, and campaign managers to produce personalisation at scale can now generate variations by the thousand. The marginal cost of a competent message approaches zero.

Personalisation becomes structural rather than exceptional. What used to require elaborate rules, segments, and manual orchestration becomes table stakes. Contextual messaging gets cheap and automatic.

Workflows and decisions accelerate. Planning, testing, audience creation, reporting, optimisation — all compress. Marketing becomes faster, more fluid, more automated, more agentic.

Interfaces change. Customers increasingly interact through conversation rather than navigation. Some decisions — discovery, comparison, even transaction — may be delegated to agents acting on behalf of humans. The human remains the principal. The agent begins to mediate more of the journey.

These are real changes. Anyone in marketing who ignores them is not being contrarian. They are being careless.

But the AI conversation has a consistent habit of confusing surface-layer change with structural-layer change.

Content creation is a surface layer. Workflow is a surface layer. Interface is a surface layer. Decision speed is a surface layer. They matter. They change. They can transform industries. But beneath them sits something more permanent: the relationship between a brand and a human.

Someone still needs to initiate contact. Attention is still finite. Trust still compounds slowly. Memory still matters. Familiarity still shapes preference. Owned channels still differ fundamentally from rented ones.

AI can accelerate production. It does not abolish the asymmetry between brands and customers. A brand still needs a way to be noticed when the customer is not actively searching. A brand still needs a way to remain present between transactions. A brand still needs a way to avoid paying again and again for attention it once owned.

That layer predates digital marketing entirely. It predates search, social, and martech. It has survived every interface change so far. And it is exactly where the largest unclaimed opportunity still sits.

This series is about that enduring layer. Not because AI is unimportant. But because the companies that build only on what changes will always be running to catch up. The companies that build on what endures get to compound.

The deepest layer in marketing is not content. It is not automation. It is not even intelligence. It is attention inside a relationship. That is where we begin.

Thinks 1951

Eric Lamarre: “One of the core questions CEOs have is: “Is this worth it? How long is it going to take?” We analyzed 20 of what I would call the best of the best companies that have executed tech and AI transformations very well. Three numbers stood out. First, they achieve about a 20 percent EBITDA uplift on average from their AI-focused efforts. And that value typically comes from two or three areas at most. There’s an extreme focus on the economic leverage points. Second, one to two years to pay back. In some instances, like Freeport, it was well below a year. On average, though, it takes one to two years before you’re in the green and then you can continue to innovate…And third, for every dollar invested, they got about three dollars of EBITDA. You don’t often see returns like that. Why do they get those returns? Because they focus on the economic leverage points, places in the business model where even a small improvement through AI drives massive value.”

WSJ (Acquired) on Ferrari: “On the surface, this strategy of scarcity and tantalizing exclusivity seems remarkably similar to the classic playbook of other storied luxury brands, as if Ferrari’s cars are just Hermès bags and Rolex watches on four wheels.  But Ferrari has something that Hermès and Rolex could never cultivate: hordes of screaming fans who worship the brand from the time they can say vroom-vroom…Despite selling a grand total of 330,000 cars over the course of its entire history, Ferrari boasts more than 400 million fans around the world. And no company has a higher ratio of people who know about its products to people who actually own those products. Far from cheapening the brand, Ferrari’s rabid base of superfans only enhances the brand’s appeal to clients who can afford to pay millions of dollars for a car they will rarely drive.”

FT: “Asia’s technology sector is undergoing a shift. The era dominated by consumer apps, from ecommerce marketplaces to ride-hailing platforms, has been giving way to a wave of start-ups focused on applying AI to established industries. Many of the fastest growing companies in the Asia-Pacific region are emerging from this shift.”

NYTimes: “In just six months, a team at Children’s Hospital of Philadelphia and Penn Medicine designed a personalized therapy that could correct the single misspelled letter in KJ’s DNA using a gene editing technology known as CRISPR. To get the therapy inside KJ’s cells, doctors relied on the same kind of mRNA technology that powered the Covid-19 vaccines…Recent advances in mRNA science and CRISPR gene editing mean that the approach that helped KJ could be used for other children. The technology can be reprogrammed for different diseases by inputting a short stretch of genetic code that tells the molecular machinery exactly where to make its correction. Build the system once, and you can redirect it to a new disease by changing that one piece.”

NeoMails: The Just A Minute Philosophy

Published May 5, 2026

In an age of infinite scroll, the most powerful relationship product may be one that earns just a minute — and makes that minute worth repeating

1

The Philosophy the Inbox Was Missing

I grew up listening to Just A Minute on BBC World Service. [Here’s a sampling.]

For those who did not, the format was deceptively simple: one contestant, one topic, sixty seconds, three rules. No hesitation. No repetition. No deviation. Nicholas Parsons presided with the calm authority of someone who had seen everything and found all of it delightful. The buzzer would interrupt, a point would be awarded, the round would end.

It finished.

That sense of completion is more important than it first appears. Much of modern digital product design is built around the opposite principle. Infinite scroll does not want to finish. Feeds do not want to end. Autoplay does not want to stop. The most successful attention products of the past fifteen years have all been engineered to remove stopping cues, elongate sessions, and convert curiosity into compulsion. Their core logic is simple: if attention is valuable, then more time spent must always be better.

Brands copied the same instincts. More emails, more nudges, more urgency, more retargeting, more frequency. The result is what younger generations now call brain rot — not a failure of content quality but a failure of architecture. An infinite feed is never finished. That incompleteness is not a side effect. It is the product. The residual unease of never having reached the end is the mechanism by which these platforms hold attention.

There is a cognitive science name for this: the Zeigarnik effect. Incomplete tasks occupy working memory; complete tasks release it. An infinite scroll produces a form of low-grade anxiety — the sense of having consumed without finishing. NeoMails are built on the opposite premise. They are not trying to capture as much time as possible. They are trying to make one minute worthwhile — and to make that minute feel finished. The mild satisfaction of having done a small thing is not incidental to habit formation. It is the mechanism of it.

The three JAM rules as design principles

The BBC programme’s three rules map, with unexpected precision, onto what a worthwhile inbox interaction requires.

No hesitation means the value has to arrive immediately. A contestant who pauses before beginning loses the point. In NeoMail terms, this is the role of the BrandBlock. The brand’s voice, context, and perspective appear first — before anything is asked of the reader. The message begins cleanly. Value delivered before anything is asked of the reader.

No repetition means familiarity without staleness. A speaker who recycles a word or phrase loses the point. For the Magnet — the interactive element at the heart of the NeoMail — repetition is the death of habit. The format must be recognisable enough to require no learning, but the content must change every time. Familiar enough to feel easy, fresh enough to feel worth returning for.

No deviation means the format must honour its own purpose. A NeoMail that opens with a discount is a Sell message in disguise. A Magnet that is really a product-preference survey with a quiz skin on top is still a brand asking for something, not giving something. The reader sees through it immediately, and the trust the format has built begins to erode.

These are not arbitrary constraints. They are the conditions under which a sixty-second inbox interaction remains worth repeating indefinitely.

The APU as the atomic unit

The APU — Attention Processing Unit — is the design architecture that makes the JAM philosophy operational. Three components, three jobs, one bounded minute.

The BrandBlock gives the NeoMail identity. The brand speaks first, in its own voice, before anything is asked. The Magnet gives it lift — the reason this email is worth opening today rather than archiving with the rest. And Mu gives it memory — the accumulation that connects today’s interaction to tomorrow’s, and tomorrow’s to the weeks that follow.

Together they create something no conventional email format has produced: an interaction with a beginning, a middle, and an end. A message you can complete.

Social platforms monetise attention by extending duration. The APU monetises attention by increasing quality and continuity per minute. These are not just different strategies. They are different philosophies of what attention is for.

The inbox does not need more content. It needs a better minute.

Just A Minute ran for decades. It outlasted a hundred trendier formats. It built a loyal audience not by demanding more time but by being reliably worth the time it asked for. One minute. Finished. Repeated. That is the model.

2

The Magnet: Design Science for the Most Important 20 Seconds

If the APU is the unit, the Magnet is the load-bearing element inside it.

BrandBlock gives the minute identity. Mu gives it memory. But without the Magnet, the NeoMail is still a better-formatted email — more thoughtful, perhaps, but still passive. Still something to consume rather than participate in. The Magnet is what changes the fundamental posture of the interaction. It converts the reader from recipient to participant.

That distinction matters because people do not return to a brand’s inbox for information. Information is available everywhere in quantities no one can process. People return for participation, anticipation, progress, and the closure of a small thing completed. The Magnet delivers all three in under sixty seconds. Understanding how requires treating Magnet design not as content creation but as a design science.

Two axes that organise the space

Before cataloguing formats, the architecture that makes sense of them. Every Magnet sits on two dimensions.

The first is what you risk: free to play, where participation earns Mu on accuracy or completion, versus pay Mu to play, where the reader stakes currency for the possibility of winning more. The free formats minimise activation energy — anyone can participate right now at no cost. The stake formats introduce commitment — the reader has skin in the game, which changes the quality of attention brought to the interaction.

The second is when you find out: instant result versus deferred result, where the reader returns tomorrow to see how a prediction settled. Instant formats close the loop immediately. Deferred formats create cross-session engagement by design — you predict today, return tomorrow.

These two axes produce a natural gradient of engagement intensity. The free, instant Magnets build the daily habit. The stake-and-wait Magnets deepen it for committed users. A well-designed NeoMail programme uses both, with the free formats dominant at onboarding and the stake formats introduced as the Mu balance gives users something worth risking.

Four families, four psychologies

Opinion and social — free, instant, earn Mu. The psychology here is self-expression. Humans have a consistent and underestimated preference for registering an opinion and discovering what others chose. A preference fork — Would you rather, Hot or Not, This or That — resolves in seconds, requires no prior knowledge, and generates the most accessible engagement of any format. The social signal that follows — most people said B; you said A — creates a mild belonging that is distinct from anything a promotional email can produce. New formats in this family extend the mechanic: Rank these four, Caption contest (pick the funniest from four options), Rate this (a product, a moment, an idea). All share the same engine: express yourself, see the crowd.

Skill and knowledge — free, instant, earn Mu on accuracy. The psychology here is the curiosity gap and the competence reward. A question creates micro-tension. The correct answer resolves it. The satisfaction is in knowing — or in discovering you were wrong in a way that is informative rather than punishing. The brand’s role is to make the topic adjacent to its world without making it a product pitch. A financial services brand quizzes on market trivia; a food brand on ingredient origins; a fashion brand on emerging designers. The key design constraint: questions that reward knowledge feel satisfying; questions requiring specialist information unavailable to most readers feel like homework. The taxonomy is rich: Trivia quiz, Emoji decode (guess a brand or film from emoji), Price is right (closest wins), Connections (odd one out), Word game (Wordle-style, four letters, three tries), True/false blitz (five rapid-fire statements), Blind brand test, Before/after reveal, Spot the difference.

Prediction — stake Mu, deferred result, reputational compounding. The psychology here is investment. Once Mu is staked on an outcome, the reader has skin in the game. This is the only family that creates cross-session engagement by design. You predict today. You return tomorrow not because the brand asks you to, but because you want to know if you were right. The Predictor Score — the compounding, portable record of forecasting calibration — is unique to this family. Over time, a high Predictor Score becomes a reputation, and reputation becomes a reason to return entirely independent of any brand agenda. Formats: WePredict teaser, Fast forecast (will X happen today?), Crowd vs expert (agree or disagree with an analyst’s call).

Games of chance — pay Mu to play, winner-takes-most, instant. The psychology here is risk and reward. Unlike Skill Magnets where accuracy earns Mu, Chance Magnets require burning Mu to play. Book cricket, scratch card, pick a door, double or nothing, horizontal roulette — these are highest-engagement for users with established Mu balances and lowest-appropriate for onboarding. They should appear later in the NeoMail journey, when the Mu balance gives the reader something worth risking.

The design discipline: familiar novelty

What separates a Magnet that sustains a daily habit from one that entertains once is a constraint that sounds simple and is genuinely hard to execute: familiar novelty. The format must be recognisable enough that no cognitive load is spent understanding how to participate. But the content must vary enough that returning feels different from yesterday.

A Wordle-style word game can run daily — the format is fixed, the word changes. A preference fork can run daily — the structure is fixed, the choices are fresh. A trivia quiz fails if the questions become predictable within the brand’s narrow topic range. The format earns the return; the content justifies it.

Magnet design is less about content creation and more about format design. Once the format is right, content can vary endlessly — and increasingly, AI can help supply that variety at scale. But no amount of content abundance compensates for a weak underlying format.

The failure modes

Three ways Magnets break. Too much friction: a Magnet that requires reading a paragraph before participating has already failed — the interaction must be legible within five seconds. Brand questionnaire in disguise: “Which of these products would you most like to see?” violates the no-deviation rule and the reader sees through it immediately, and trust decays. Repetition without variety: the same question structure week after week with cosmetic variation leads to category fatigue far faster than brands expect.

The Magnet is not decoration inside the APU. It is the engine. The BrandBlock earns the brand’s presence once the reader is already participating. Mu earns tomorrow’s return. The Magnet earns this open — and earns the right for everything else to follow.

Design it well and the minute is worth repeating. Design it poorly and no amount of Mu will compensate.

3

BrandBlock and Mu: How One Minute Becomes a Relationship

A good Magnet can make a NeoMail interesting. But one interesting interaction is not yet a relationship. The reason the APU matters is that the other two elements — BrandBlock and Mu — transform an engaging minute into something that compounds over weeks and months.

The Magnet creates participation. The BrandBlock makes that participation belong to the brand. Mu ensures that today is not disconnected from yesterday. That is how one minute becomes a relationship.

BrandBlock: the brand earns the right to be present

The Magnet produces a rare state in the inbox: an attentive, participating reader. That is the precise moment the brand inherits. In conventional email, the brand tries to force attention first and hopes for participation later. In a NeoMail, the sequence reverses. Participation comes first, and the BrandBlock benefits from the reader’s activated state.

That makes the BrandBlock strategically important. It is not an ad slot. It is not filler. It is where the brand’s world is expressed — voice, perspective, category point of view, product context, the small signals that make this minute belong specifically to this brand and not some generic engagement machine.

A fashion brand may frame the season or style mood. A financial services brand may place a market moment in context. A food brand may tell a brief origin story. A beauty brand may spotlight an ingredient or technique. The BrandBlock does not have to sell to be commercially useful. In fact, its usefulness depends on not selling too directly. The customer has just participated. The brand’s job at that moment is to convert participation into familiarity — not to interrupt it with an offer.

The design constraint is strict: the BrandBlock cannot carry a promotional offer. A NeoMail that opens with a discount is a Sell message in disguise. The trust built over weeks of consistent Relate is not instantly destroyed by one violation — but it is nicked, and nicks compound.

Mu: the visible memory of the relationship

Every conventional email arrives from nowhere. It does not know about the one before. It does not acknowledge the reader’s history with the brand. The backend may remember everything. But from the customer’s point of view, the inbox has no memory — each send is a stranger introducing itself, regardless of how many times the introduction has been made.

Mu changes this structurally.

But Mu should not be mistaken for a conventional loyalty programme. The distinction is important enough to state plainly.

Loyalty programmes reward spend. Mu rewards showing up. Loyalty points usually sit in the background until redemption. Mu sits in the foreground as continuity. Loyalty is about delayed transaction incentive. Mu is about making the relationship itself feel cumulative.

The Mu count visible in the subject line tells the reader, before the email is opened, that yesterday mattered. That showing up left a trace. That this interaction is part of something larger than a one-off message. A Mu balance built over weeks represents something real: time given, decisions made, habits maintained. That accumulated weight makes the next open more likely — not because the reader is chasing points, but because the relationship has a record, and records feel worth continuing.

Mu also functions, in ways conventional martech cannot replicate, as a leading indicator of attention decay. A falling Mu balance — a slower earn rate, a broken streak — predicts drift before open rate does. Open rate is binary: the email was opened or it was not. Mu velocity measures the quality and consistency of engagement over time. A brand monitoring Mu balances across its Rest segment has an early warning system that no campaign dashboard provides.

Earn vs stake: how Mu creates two kinds of engagement

Free Magnets earn Mu — participation is rewarded with accumulation. This is the daily habit mechanism. Low friction, immediate reward, easy to repeat.

Chance Magnets require staking Mu. This deepens the system. The customer is no longer just collecting; they are committing. Once Mu can be risked — in prediction or games of chance — it begins to feel more alive. Not because it has become money, but because it now shapes behaviour in two directions: earning and burning.

A NeoMail programme that only offers free Magnets is sustainable but shallow. The Mu balance grows but never feels truly valuable because it is never at risk. One that introduces stake-based experiences at the right point — after the habit is established, after the balance is meaningful — creates the intensity of engagement that makes Mu feel like a real currency rather than a decorative counter.

How one minute compounds into a relationship

Day one: the reader opens out of curiosity. The Magnet is quick. The BrandBlock is noted. Mu appears in the next subject line. Week two: the format requires no learning to enter. The streak is visible. The activation energy of opening has fallen. Month two: the brand is part of a weekly inbox pattern. The BrandBlock is absorbed by a reader already in motion. Month six: the Mu balance represents genuine attention investment. The Predictor Score has a history. The relationship has a record — and records are harder to abandon than novelties.

No conventional email programme creates this arc. The APU is the only inbox format designed for accumulation, and accumulation is what transforms a sequence of interactions into a relationship.

The commercial implication

The APU does not replace Sell and Notify. It earns the right for them to be heard. A customer who has spent three months in a light NeoMail rhythm is different from one who receives only promotions and transactional messages. The former has continuity. The latter has interruption. When the time comes for a launch, a replenishment reminder, or an ActionAd, the brand speaks into a relationship that has stayed warm rather than trying to restart one from scratch.

This may be the deepest difference between infinite-scroll systems and APU systems.

Feeds maximise duration. APUs maximise recurrence with closure. Feeds want you to stay. APUs want you to return.

For brands, return may ultimately matter more than duration. And return, built on the foundation of a completed minute repeated over months, is what the APU is specifically designed to create.

Nicholas Parsons once observed that Just A Minute worked because the constraints forced a quality of attention that open-ended formats never required. The speaker had to be genuinely present. No hesitation, no repetition, no deviation — not arbitrary obstacles, but the exact conditions that made the minute worth the attention it asked for.

The APU imposes the same discipline on the inbox. The BrandBlock must earn its presence. The Magnet must be worth completing. Mu must mean something because it accumulates from choices that cost time.

The inbox does not need endless content. It needs the right minute, every time. Bounded, worthwhile, and repeated with memory until the minute becomes a relationship.

That is the Just A Minute philosophy. And that is what the APU is built to deliver.

Thinks 1950

Christoph Schweizer (BCG newsletter): “AI is more likely to amplify the roles of software engineers than eliminate them. Human judgment still matters in system design, architectural tradeoffs, quality assurance, and integration. And, as the cost of building software falls, demand can expand to meet unmet needs for digital products, automation, and new features. Obviously, this dynamic could change as AI models mature. Much of the work of call center representatives is structured and repeatable, while demand is largely fixed. The volume of incoming calls is unlikely to increase sharply because wait times go down. In those settings, AI is more likely to substitute directly for labor in all but the most complex cases.”

FT: “The rise of Taiwan’s chip industry is one of the most remarkable industrial stories of our century. But the island of 23mn people lies on a geostrategic — as well as a seismological — faultline, roughly 100 miles off the coast of China. Beijing has long trumpeted its “national rejuvenation” mission to incorporate the island. It has also significantly boosted its military capabilities to help achieve that end. Any serious disruption to the global supply of the world’s most valuable semiconductors would surely bring the current AI investment boom in the US screeching to a halt. It would also rattle global stock markets that are heavily leveraged on Big Tech’s colossal AI bet. The over-reliance of the US on Taiwan’s manufacturing output has been belatedly recognised by Washington, but overcoming that challenge is an altogether different proposition.”

The Generalist: “Confidential is a wildly entertaining and impressively insightful book. In studying it closely these last few months, I’ve also come to believe it’s an important one. Though [Jeff] Nolan is ostensibly writing for the professional intelligence gatherer, his conversational techniques are useful to anyone, in any context. They are liable to make you more engaging and persuasive, as well as a better conversationalist. It is also worth knowing when someone else is using them. Why did that salesperson seem to purposefully misspeak? Was I imagining it, or did that headhunter seem to disbelieve everything I said? What is it about this person that makes me want to open up so much? For founders working in sectors of national interest, Confidential will help you protect what you know. If you are building almost anything of note, there is a good chance that someone out there — whether in a bland concrete building, a glassy office tower, or a grassy tech campus — would love to understand it better than you’d like them to.”

Adrian Wooldridge: “There is one thing you can do to ward off existential despair. Go to your local coffee shop and order a cup of coffee. Not only is relaxing over a cup of coffee a perfect therapy in troubled times. The world’s booming coffee culture is a sign of the health of the liberal order.”