Inverting the interface: design, personal agents, and the post-brand world
What happens when people experience brands, services, and institutions primarily through their own AI agents—rather than through brand‑designed apps, websites, and campaigns?
The speculative question is simple (and slightly annoying, if you make it real):
What happens when people experience brands, services, and institutions primarily through their own AI agents—rather than through brand‑designed apps, websites, and campaigns?
Not “agentic AI” as enterprise workflow automation.
Agents as personal infrastructure. A user-side layer that sits on top, negotiates, filters, and renders the world—on the user’s terms.
Under that scenario, two things happen at once:
- Design moves away from screens.
- A comfortable industry myth gets stress-tested: “we design for the user” (and its sibling, “we put the customer first”).
Because structurally—most of the time—we didn’t. We designed and branded for the business via the user.
Download the full report (PDF): Inverting the interface.
~30 min read · PDF · no forms, no tracking
Disclaimer
This post is about a speculative scenario report on loyalty in times of agentic AI, not a forecast.
It’s designed to sharpen strategic conversation—not to predict a single future.
This post was not generated by AI.
Generative AI tools were used selectively as an assistive drafting and editing aid—to accelerate outlining, improve clarity, tighten language, and generate alternative formulations and illustrative examples.
As a practical estimate of contribution at the level of ideas and intellectual ownership, roughly ~75% of the thinking, structure, and argumentation originates from the author, while ~25% reflects AI-assisted drafting that has been reviewed, edited, and integrated by the author.
The author remains solely responsible for the final content, including interpretation, emphasis, and any errors or omissions.
All data from publicly available sources.
Any brands, trademarks, or organisations referenced are used purely as illustrative examples. This paper has not been commissioned, endorsed, or sponsored by any of the brands mentioned; at the time of writing I had no commercial, advisory, or employment relationship with those brands.
All opinions expressed in this article are solely my own and do not represent the views of any current or former employer.
Content
- i. The inversion
- ii. Who owns the chrome, who owns the agent
- iii. From SEO tricks to agent alignment
- iv. What shifts for UX, UI, and brand
- v. The uncomfortable questions
- vi. What changes for brands, retailers, and retail media
i. The inversion
The backbone idea is a reversal of the last ~25 years of digital interface logic:
- Today: brands design an interface → users adapt to it
- Speculative future: users (and their agent) define an interface → brands adapt to that
This isn’t Minority Report theatre. The “display” can be anything: phone, voice, AR, ambient surfaces, micro-interfaces across devices.
The point is structural:
A stable, user-centric experience layer sits above many brands and services, and that layer is mediated by the user’s own agent.
In other words: the primary interface in your life is not your bank’s app. It’s your agent’s window onto your bank.
ii. Who owns the chrome, who owns the agent
The less glamorous layer matters: chrome.
Chrome is the framing interface around content: browser window and tabs, OS status bars and notifications, system gestures, wake words, default response patterns.
For decades, whoever owned the chrome wrote the rules of experience. You could build anything you wanted—as long as it lived inside someone else’s rectangle, patterns, and policies.
The provocation here is not “custom dashboards”.
It’s:
chrome becomes user-owned (or at least user-defined) once an agent sits between brands and the person.
If the agent is the arbiter of density, hierarchy, modality (text/voice/visuals/haptics), allowed patterns, and attention rules—then “owning the chrome” becomes a subset of owning the agent.
And that takes us to a layered ecosystem:
- Foundational agent: your constitutional layer
(identity, consent, values, accessibility needs, long-term constraints) - Vertical agents: domain specialists
(travel, shopping, money, health, public services), operating under foundational rules - Branded agents: guests run by organisations
(your bank, Amazon, Zara, tax authority), filtered and vetoed upstream
Design and power concentrate at the top.
Not in the branded agents. Not in the brand app. At the foundational layer.
iii. From SEO tricks to agent alignment
There’s a cousin to this whole story: SEO.
SEO existed because:
- users expressed vague desires in a text box
- search engines had crude models of intent
- an industry emerged to game those proxies
Design has been doing something eerily similar.
Call it internal SEO for UX:
- interpret messy “user needs” inside business constraints
- translate them into flows
- optimise for proxy metrics
(clicks, NPS, conversion) - sometimes helpful, often funnel-theatre
In an agent-mediated world, brands still try to optimise for agents (of course they will). “Agent Experience Optimisation” will be someone’s new job title.
But the rules change.
If the user’s foundational agent is genuinely aligned with the user’s interests, the old tricks stop working. You don’t get better placement by shouting louder or hiding fees.
You get it by being:
- semantically clean
(agents can understand you) - behaviourally reliable
(you do what you say) - policy-compatible
(you respect explicit user rules)
That’s not branding as story. That’s branding as behaviour under machine scrutiny.
iv. What shifts for UX, UI, and brand
From this vantage point, four shifts compound.
1 - Design moves to the agent layer
The most consequential design decisions no longer live in screens and campaigns, but in how agents see, interpret, and act:
- objectives
- semantics
- protocols
- behaviours
- negotiation rules
A large share of what we currently call UX/UI (templated flows, standard patterns, funnel tweaks) becomes automatable—because it already is codified.
2 - Brands lose control of the interface (and will fight back)
When user-side agents strip away chrome, brands risk being reduced to what they can express as:
- data
- constraints
- guarantees
- policies
Some brands adapt and become protocol-native.
Others double down on walled gardens and hostile interfaces. Expect legal and technical resistance. Expect platform brands to try to become the vertical agents themselves.
3 - UX shifts to semantics, protocols, information architecture
UX becomes closer to:
- data architecture
(content models, entities, events) - experience protocols
(capabilities, sequencing, fallback, escalation, consent) - governance of behaviour
(auditability, explainability, constraints regulators can interrogate)
Still design—but the material is semantics, not buttons.
If you want the blunt version:
the job shifts from “SEO for screens” to meaning-making for agents.
4 - UI design collapses into design-system stewardship
UI doesn’t disappear. Its centre of gravity changes.
With mature design systems + tokens + agents that can compose UI, a lot of UI work becomes:
- maintaining systems
- enforcing consistency and accessibility
- evolving a visual ontology
(“what does a warning look like?”, “how does urgency map to motion or sound?”) - curating quality across agent-generated variants
And if agents go multimodal (they will), today’s screen-trained interface practice looks… narrow.
v. The uncomfortable questions
This is where the report stops being polite.
The “user-centred” story has often been structurally compromised.
- UX teams sat under growth/product/marketing → KPIs were conversion, retention, engagement, NPS.
- Brand teams sat under marketing/comms → KPIs were awareness, consideration, preference, share of wallet.
The slogan on the wall said “for the user” the way the plaque in a corner shop says “the customer is always right”:
rhetorically about the human, structurally about the till.
In that light:
- free returns and same-day delivery aren’t generosity; they’re extraction with deferred costs
- frictionless betting isn’t “engagement”; it’s a funnel into addiction
- cross-selling journeys aren’t “helping you discover value”; they’re mining attention and weakness
Designers weren’t always calling those shots.
But they were almost always the ones translating those shots into pixels, flows, and narratives—and then calling it “user-centred” because it felt smooth.
Personal agents introduce, for the first time, a plausible technical path for “for the user” to stop being a slogan and start being an architecture:
Either the user owns the foundational agent that mediates the world for them, or they don’t. Either brands can express value in clean protocols agents can trust, or they can’t.
So the questions land where they should:
As a designer: are you investing in the skills that help businesses decorate other people’s agents, or the skills that define what those agents are allowed to see, do, and optimise for?
As a brand: are you still spending most of your energy on make-up on the chrome, or are you investing in systems and semantics that survive when the chrome, the app, and even the human-designed UX are gone?
Download the full report (PDF): Inverting the interface.
~30 min read · PDF · no forms, no tracking
✺
Want to go deeper?
Strategy Activation Lab
A focused executive working session designed to turn the findings of this speculative scenario into a clear set of insights, priorities, and next steps—such as:
- a map of your loyalty value layers as machines would compute them
- “agent-ready” program interfaces and governance
- capability gaps across data, operations, product, and design
- a short list of moves you can make in the next 90 days
This is not a creativity workshop with post-its, but a structured, outcome-driven lab that aligns stakeholders around a shared map of the world.