Designing places for the invisible user?

When AI agents navigate our cities

Walk down London’s Oxford Street in ten years’ time, and the crowd might look the same, but no two people will be seeing the same city. Through a pair of augmented lenses, your personal AI knows what you like: independent coffee, quiet side streets, sustainable fabrics, size seven shoes. As you walk, it quietly edits your field of vision, highlighting a tucked-away gallery, suggesting a detour through a calmer street, reminding you of a brand that matches your values. The city rearranges itself around you.

Now let’s try and imagine everyone else’s version of Oxford Street. Each one different. Each one curated by a digital companion that knows its human better than they know themselves.

For decades, we’ve talked about human-centred design. But what happens when the user isn’t really human anymore? When our experiences of the city are increasingly mediated through artificial intelligence? Are we, as built-environment professionals, ready to design not just for people, but for the AI agents that guide them?

From smart cities to sentient streets

The “smart city” movement promised seamless efficiency: sensors in lampposts, responsive lighting, data-driven planning. The next phase will be more intimate and more invisible.

AI agents - personal, portable, always-on - will soon be the filters through which most people experience place. The question isn’t just how the city uses data to serve us, but how our data companions use the city to serve us. Where we once designed legibility for the human eye, we may soon be designing legibility for the “Algorithmic One”. What information will a building need to “broadcast” for an AI to know it’s relevant to its user?

In marketing and retail, this shift is already well-underway. AI systems are learning to analyse sales calls and predict footfall. The integration of data and narrative is becoming the new craft.

The built environment will be next.

A new stakeholder in the built environment

Developers already think in layers of audience: investors, tenants, residents, visitors. Soon, there will be a fifth: the AI intermediaries acting on behalf of all the others. An AI doesn’t stroll down the high street or admire façades. It scans, parses, and prioritises. It will seek metadata, not mood…. At least for now. Its choices will be guidedby structured information: accessibility data, energy efficiency scores, digital ratings, pricing algorithms. A building that fails to surface this data clearly may become invisible to the digital agents that recommend where we eat, shop, or stay. Places could rise or fall in prominence based on how “discoverable” they are to machines.

Urban visibility will have a second dimension: search visibility. Think of how websites are designed today. The language of Search Engine Optimisation (SEO) like tags, keywords, quietly dictates how content is built. Tomorrow’s architects and planners may face a similar challenge: creating cities that are both humanly rich and algorithmically legible.

Maybe what we need is a new term: Physical Experience Optimisation: PEO, designing for the “visible engine” rather than the search engine.

Personalised cities vs shared experiences

The promise of AI-mediated living is hyper-personalisation: places and experiences tailored precisely to your taste and mood. But there’s a quiet paradox here. When every journey is personalised, the collective experience of the city begins to fragment.

If each citizen walks a different route, guided by an invisible algorithm, what happens to the serendipity of stumbling upon a street musician or a local market? There’s also the risk of flattening. AI systems, trained on preference and pattern, may start steering us toward what’s “most likely” to please us, which means places begin to optimise for the predictable. When AI optimises for efficiency, it erases nuance and distinctiveness.

The intangible, the slightly offbeat, the spaces that rely on a sense of vibe or gut feel could struggle to compete in a world where every decision is data-rationalised. In the process, we might lose something deeply human: the conviction that a place moves us for reasons we can’t quite name. And beneath that sits a subtler danger: the creation of physical echo chambers.

If AI agents continuously curate our routes and routines to align with our interests and comfort zones, our cities could begin to mirror the social media feeds we already inhabit: self-confirming, familiar, homogenous.

Imagine a world where sports fans, foodies, and luxury shoppers all move through separate invisible geographies, rarely overlapping. The public realm, long a space of friction and encounter, risks splintering into micro-realities that never quite meet: a mosaic of overlapping privacies.

Storytelling for machines

In marketing, we’ve long known that data without narrative doesn’t move people. But as AI begins to mediate choice, we’ll need stories that machines can understand. Every building, brand, and district will require a kind of digital mirror: a structured narrative layer that tells AI agents what it stands for, who it serves, and why it matters.

For property and place brands, this isn’t science fiction, it’s a new literacy. Imagine a future where an AI assistant curates your day based not on location alone, but on alignment with your values: inclusive design, cultural authenticity, walkability.

The future won’t just reward visibility; it will reward codification. Places that can clearly express their qualities, eg inclusivity, cultural authenticity, sustainability, walkability; through consistent, machine-readable data will be the ones surfaced first.

Codification becomes the new storytelling. Not in the sense of reducing meaning to metrics, but of translating meaning into metadata. The urban storytellers of tomorrow may need to write in two languages: human and machine.

Early signals: the algorithmic city is already here

This future isn’t entirely hypothetical.

In Copenhagen, real-time mobility data already adjusts traffic lights and bike routes dynamically. Helsinki has trialled AI-driven public-transport optimisation. Paris uses predictive models to plan cultural programming based on crowd sentiment.

Further afield, Seoul and Singapore are at the frontier of adaptive urban systems with AI-driven mobility systems. Shopping malls in Dubai use predictive analytics to direct visitors to quieter routes or higher-margin retailers. In Tokyo, digital twin districts simulate how people and data interact, allowing developers to test scenarios before construction begins.

These examples show the early outlines of an “algorithmic urbanism”, where design, data, and behaviour feed each other continuously. But they also raise difficult questions: who controls the code that shapes our perception of space? And how do we ensure that AI augments, rather than replaces, human curiosity?

Ethics, agency, and the loss of serendipity

AI could help design more responsive, inclusive environments, ones that adapt to individual needs, guide vulnerable users, or balance real-time demand. But it could also narrow our worlds, turning cities into comfort zones optimised for convenience. When the algorithm knows your preferences too well, it stops showing you anything new.

For urbanists, that’s a profound challenge. Cities have always thrived on friction: the unexpected encounter, the accidental view, the tension between order and chaos. If we outsource our navigation, discovery, and even aesthetic judgment to AI, we risk sanitising the very messiness that makes urban life meaningful.Physical echo chambers would be the ultimate consequence: invisible partitions separating communities not by class or race, but by algorithmic curation. If left unchecked, this could quietly erode civic empathy: the simple understanding that others move through the same spaces differently.

On a human level, what will this do to us cognitively if we only ever encounter the predictable and the known? The city has always trained our brains to be alert, adaptive, improvisational. If AI filters out surprise, we risk a kind of perceptual atrophy.

And underpinning all of this is a question of trust. As AI mediates more of our choices, we may grow less confident in our own instincts and more dependent on systems whose logic we can’t fully see.

Perhaps our job in the AI age won’t be to perfect the user experience, but to protect its imperfections to design for wonder as much as for efficiency.

Reclaiming the human

AI will soon shape how we see, interpret, and value the world around us. But cities are not spreadsheets, their worth lies as much in atmosphere as in analytics; in the way light hits stone, or laughter echoes through a square.

As we begin designing for the invisible users, the AI agents who will navigate on our behalf, we must resist the urge to make our environments too knowable, too optimised.

The next era of urbanism will belong to those who can balance machine logic with human intuition. While AI might be able to tell us where to go, only we can decide why it’s worth the journey.