Welcome to this week’s edition.
We are currently living through a bizarre, split-screen moment in tech and business.
On one side of the screen, Silicon Valley is double-downing on an unprecedented, compute-heavy infrastructure built on top of a fundamentally inefficient software engine.
But on the other, this very brute force is quietly rewriting the entire relationship between brands, interfaces, and human autonomy.
This week, we’re unpacking the systemic shifts happening right under our noses: from the illusion of AI scaling laws to the active outsourcing of our cognitive perimeters. We’ll look at how the traditional search box is being replaced by autonomous agent farms, why personalised synthetic media is shattering our shared cultural realities, and how "algorithmic loyalty decay" is rendering traditional marketing funnels obsolete. Finally, we’ll look at the counter-intuitive category design plays being run by the world’s dominant Category Kings—and how you can apply their scoreboards to protect your own business model before the ecosystem tips permanently.
Let’s dive in.
We’ve built an economic and technological consensus around a foundational lie: that the “large” in Large Language Model is a strategic feature, not a structural failure.
The tech industry has successfully marketed a massive engineering limitation as if it were a badge of honour.
“Large” in Large Language Model, isn’t the “large” cargo ship or “large” factory - a sign of power, capability, and economies of scale. In computer science, a system that requires hundreds of billions or trillions of parameters just to perform basic language tasks isn’t efficient; it’s structurally bloated.
Here's why this “large” is a failure, not a feature:
The Quadratic Trap: Traditional, elegant software scales logarithmically—meaning the bigger the task, the more efficient the software gets at handling each additional piece of data. LLMs scale quadratically. As you feed them more tokens to process, they slow down and consume memory at an exponentially compounding rate.
The Diminishing Returns Problem: Because the underlying architecture is so inefficient, companies are trapped on a treadmill. To get a minor, incremental improvement in capability, they have to expand the model size at a massive, unsustainable rate. It’s the engineering equivalent of adding fifty exhaust pipes to a car just to get two extra horsepower.
The Brute-Force Bubble: Tech giants are throwing billions of dollars at data centres and gigawatts of compute because it is a low-risk, predictable way to invest capital right now. It’s far easier to throw money at a known, bloated architecture than to pause and do the hard research required to build something fundamentally elegant, small, and efficient.
As “personal intelligence” features deploy globally, with Google expanding ambient AI features across nearly 200 countries and 98 languages, we may becrisis of cognitive autonomy.
The pitch is seductive: hook up your Gmail, Photos, Calendar, and your Docs, and let an ambient information agent run 24/7 in the background, anticipating your needs and orchestrating your life.
But for those designing brand and customer experience strategies, this isn’t just a utility play. It is the outsourcing of the human cognitive perimeter.
When an interface intermediates not just tasks, but the literal texture of your day-to-day existence, what happens to human agency?
If the system knows your context, your historical behaviours, and your emotional triggers better than, say, your bank - and especially yourself - the line between anticipation and manipulation can dissolve. The same ambient agent booking your dog walker can also effortlessly infer exactly when you are most vulnerable to doomscrolling or impulsive retail therapy.
We’re moving away from traditional user experience and entering the era of decision orchestration.
Our human role is rapidly shifting from active doing to passive supervising.
We’re being reduced to a stack of “Are you sure?” prompts, clicking “approve” on choices that were algorithmically synthesised long before they reached our screens. For strategists, the challenge of the next decade isn’t building a smoother conversion flow, but figuring out how to maintain human dignity and cognitive sovereignty when context itself has been fully commoditised.
As generative AI engines increasingly remix news, opinion, and cultural content into hyper-personalised, algo-synthesised streams, the traditional concept of a unified “public” has broken down.
There is no longer a singular, cohesive audience. We’re all talking to a fragmented, entangled micro-publics, where different segments of society live in entirely distinct, AI-curated versions of reality. This personalisation isolates consumers, creating a deep echo chamber effect that fundamentally undermines collective societal trust.
When truth itself becomes highly contextualised by an interface, “Truth UX” and “Trust UX” is the most critical design surfaces an organisation owns.
If a user’s assistant can alter the narrative tone of a news story or product description based on inferred political or cultural preferences, the interface becomes a funhouse mirror. This has sparked a fascinating counter-cultural resistance: a rise in adversarial design, including “AI-proof fonts” explicitly engineered to baffle machine-vision parsers while remaining legible to human eyes.
Brands are caught in a bizarre split-screen: making their assets highly machine-readable to be indexed by agents, while simultaneously proving to hyper-sceptical human users that their information is authentic, unmanipulated, and anchored to reality.
I think the definitive corporate crisis of the agentic era will be algorithmic loyalty decay.
For decades, we’ve spent billions building emotional equity with human consumers through storytelling, brand films, and experiential marketing. But as decision orchestration takes hold, traditional feature-led differentiation and emotional brand affinity are being bypassed entirely.
If an AI agent is the entity evaluating products, choosing service providers, and executing transactions based on systemic parameters like delivery speed, refund policies, and structured metadata, traditional loyalty completely evaporates.
Your most valuable consumer is no longer only a human being navigating an emotional journey; it can be a cluster of booking and procurement bots operating completely immune to your advertising campaign.
HEALTH WARNING: Loyalty is shifting from the brand to the orchestrating interface. To survive this decay, move your differentiation deeper into the architecture of the experience. Winning requires designing an ecosystem so reliable, predictable, and structurally superior that an optimisation bot cannot afford to pass it over. If your interface can’t articulately communicate its objective value to another machine when no human is looking, your brand risks becoming entirely invisible.
How do you build a dominant company in an era where trust is disintegrating and algorithmic loyalty is decaying?
You look at the play Tesla is running.
Tesla recently published three core privacy promises: give customers choices, maintain transparency, and safeguard data. It’s an incredibly bold move for a company whose core product is essentially an ambient surveillance apparatus equipped with eight exterior cameras, cabin microphones, and a cellular link that has been phoning home since 2012.
This is trust plus experience leveraged as a category multiplier.
Apple spent a decade transforming privacy into a huge competitive moat. They didn’t sell it as a line-item, they cultivated the deeply held belief that when things go sideways, Apple protects you.
Now, Tesla is running the same play, but with an exponentially higher-stakes product that tracks your location every single mile.
They can make this promise because of a fundamental alignment of incentives. To quote Joe Pine: “If you don’t pay for the product, you are the product.” This is the absolute Achilles’ heel for foundational AI companies like OpenAI and Anthropic. When your business model relies on monetisation structures that turn user data into the product, you cannot convincingly promise protection. You can’t ask the turkey to trust Xmas. Tesla can promise privacy because they make their margin on the hardware and the software ecosystem, not on selling your driving destination history.
Furthermore, Category Kings don’t inherit old scoreboards—they invent entirely new metrics. Tesla prominently reports that its fleet helped customers avoid 37 million metric tons of CO2. Does this immediately move their quarterly P&L? No. But it dictates the rules of engagement. They created a proprietary metric, built a scoreboard around it, crowned themselves the winner, and forced legacy automakers to awkwardly react to a game they aren’t equipped to play.
Credit to Category Pirates (paywalled)
True category leadership isn’t achieved by merely shipping incrementally better features; it requires a radical commitment to expanding human flourishing.
It’s the POV that takes you and your customers into a shared near future.
Look at Tesla again. : Tesla’s airbags can deploy up to 70 milliseconds before an impact. A standard automotive airbag typically fires about 50 milliseconds after a collision has already occurred.
At motorway speeds, that 70-millisecond head start equals roughly six feet of road —the life-altering difference between your head hitting a fully inflated cushion versus a bag that is still unfolding.
Cutting that impact force by 25% will translate to many lives saved.
But this isn’t just about safety engineering. It’s a calculated, missionary stance designed to systematically dismantle the single greatest friction point holding back Tesla’s ultimate objective: Full Self-Driving and the Robotaxi ecosystem.
The primary barrier to the mass adoption of autonomous vehicles isn’t technical; it’s psychological. Humans do not yet trust the robot to drive.
So every safety proof point Tesla ships is a brick in a wall of institutional trust, closing the perception gap. The moment autonomous driving becomes provably, mathematically safer than human driving, the entire category tips permanently.
Want to test these dynamics on your own business? Run these two diagnostic plays:
Audit your give-to-get incentive structures. Take a hard look at your current business model. Where is your customer paying you with their personal data instead of their dollars? If you are relying on data-harvesting to subsidise your growth, you are exposed to a massive trust deficit. Can you re-engineer your margins so you can authentically promise absolute privacy?
Identify the core trust gap blocking your mass adoption. What’s the emotional or psychological reframe currently standing between your product and widespread market adoption? For autonomous vehicles, it’s “I don’t trust the machine to protect my life.” What’s the unspoken barrier for your customers, and what undeniable, empirical safety proof point can you create/offer to close that gap forever?
I’m Michael Cooper, and I think about entanglement - our identities, using personal agents, AI, Intelligent Interfaces, culture - and their impact on our behaviours, choice making, empowerment, and brand engagement.
If you like this newsletter and want more:
My private work, where I share original thinking with brands, strategists, and futurists, and have provocative conversations.
My LinkedIn, where I post my ideas.
I speak at conferences or in-house company events, or to enterprise teams, and anywhere when I’m allowed to ‘think aloud’ about my ideas to audiences curious about the entanglement of our humanity, culture, strategy, tech, brands and the future.
I’m currently taking bookings for late November and December.
Thanks for reading,
Michael
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.