This year at the Venice Architecture Biennale, artificial intelligence doesn’t lurk in unattended corners with its cables crossed. From powering several displays to being them, from summarising exhibition texts to wayfinding, AI has slipped from invisible, backstage logistics into the exhibition’s very scaffolding. Leading the charge here—along with curator Carlo Ratti's steadfast belief in this hybrid future—is Spatial Intelligens, an ‘AI-powered digital twin’ developed by Berlin-based architecture and design studio sub, led by Niklas Bildstein Zaar. Commissioned to design the Curator's exhibition at the Arsenale, sub has formulated a physical system of displays and circulation based on triangulation to navigate the installation, in tandem with a digital system that looks back, listens and learns, shaped by visitors’ movements, interests and interactions.

Most exhibits at the Curator's Exhibition in the ongoing biennale in Venice, Italy— running until November 23 this year—are accompanied by the usual armful of wall and panel texts: earnest, didactic, often eloquent, but lengthy. Alongside them are bite-sized AI-generated captions, pithy and pleasingly digestible. For some visitors, the latter became an increasingly welcome shorthand. Others can now follow the breadcrumbs into deeper rabbit holes, guided by the conversational AI interface offered by Spatial Intelligens.

A photograph of the exhibition is enough for this system to localise itself, to activate the show’s digital guide on visitors’ phones. Once activated, the guide both informs and remembers. Visitor interests are tracked, mapped and converted into personalised trails. At the end: a souvenir, a keepsake if you may, of behavioural patterns and personalised information for the user. “For the first time, the exhibition space itself will learn from its visitors,” the release proclaims. This seems to suggest a new operational logic for exhibition-making: one in which the curatorial is computational, the spatial is sentient and spectatorship, mutual.

Spatial Intelligens, accessible via mobile browsers, may be one of Biennale Architettura 2025’s quietest armed exercises in tracking unique behavioural profiles. The system informs, notices, reflects, retells. Architecture, thus, no longer performs or is presented in or as inert frames. It is the respondent. “Spatial Intelligens encapsulates the spirit of the Biennale: architecture as a living, evolving system of intelligence, consciousness of materials and collective experience,” they add.

Conceived closely with Ratti and his team, the complementary physical exhibition design by sub (known for their work for Balenciaga, Travis Scott, Ye and Anne Imhof), itself leans into the cybernetic to begin with. Approached and structured as a ‘nodal network’, it features a modular column grid in recycled aluminium and 3D-printed wood composites that rhythmically echo the Corderie’s historic volumes. These columns support, guide and filter, housing the 250+ installations positing an inquiry into the convergence of natural, artificial and collective intelligence in architecture and the built environment.

“This is a project that redefines how exhibitions can think, feel and respond,” Zaar notes in the official release. “The Biennale becomes a living lab,” says Ratti, “a place not for displaying answers, but for asking new questions.”

The 2025 Biennale exhibition design explores intelligence through the lens of cybernetics. The core concept envisions the exhibition as an interconnected network of nodes, where each element interacts with others to create a complex, adaptive system. – Niklas Bildstein Zaar, co-founder, sub

Considering the closure of the Central Pavilion in the Giardini this year, the exhibition's physical proponent and its design was a particularly sizeable task, with several visitors already noting how the Curator's exhibition—housed entirely in the Corderie as a result—was packed to the brim. STIR's own review of the Biennale especially commented on how a sense of being overwhelmed with the quantum of displays, decidedly tech-driven, was a common feeling for a number of visitors during the Vernissage. In that light, the launch of sub's AI proponent for the exhibition poses an especially interesting paradigm in the face of all the existing and emergent discourse surrounding this year's Biennale.

In the interview that follows, we questioned Zaar about sub’s interventions, which combine physical structures with digital interfaces to transform the conventional exhibition format into a hybrid, interactive dialogue between architecture and the audience. We also discuss the exhibition’s design as a feedback loop and a ‘learning organism’, where AI doesn’t disrupt so much as fold itself into the space, quietly, cleverly, and how that may be perceived.

Jincy Iype: How does the spatial system designed by you for the 2025 Biennale Architettura’s main exhibition at the Arsenale respond to the curatorial theme, Intelligens?

Niklas Bildstein Zaar: [It] emerged from thinking about how knowledge is organised in AI, particularly the idea of latent space, a hidden structure where meaning is stored in abstract, often unseeable ways. It felt almost magical to me and I wanted to give that invisible dimension a physical form, something that could bridge the architectural and digital worlds.

The 2025 Biennale exhibition design explores intelligence through the lens of cybernetics. The core concept envisions the exhibition as an interconnected network of nodes, where each element interacts with others to create a complex, adaptive system. The nodes exist at multiple scales, within the structural system that organises the space and as the individual exhibits. Building on this concept, we integrate physical and digital dimensions through Spatial Intelligens, our AI-powered guide that creates a cognitive layer, a responsive interface between visitors and the exhibition. Our design prioritises navigating, recognising, interacting. With Ratti’s curatorial vision, we saw this as an opportunity to explore how intelligence can become visible, even tangible, through architecture.

Jincy: What kind of research informed your core concept and idea for this?

Niklas: Our starting point was an investigation into how human, artificial and collective intelligence can be spatially represented. We looked at cybernetic models, systems theory and ideas from mathematics and AI, all of which deal with systems beneath the surface, invisible yet structured. This fascination with hidden organisational principles became central to our approach and our design concept translates these abstract frameworks into physical dimensions.

We also studied visitor behaviour and cognition. How can we design a system that doesn't just present knowledge but organises it in a way that adapts to the viewer? Our aim was to create an exhibition that maps and responds to the complexity of thinking, both individual and shared.

Jincy: How does your AI–powered system create an adaptive space that "evolves" based on visitor interactions? What specific parameters does it track to ensure a personalised experience?

Niklas: Spatial Intelligens is at the core of this adaptive experience. Our AI-powered guide creates a cognitive layer, establishing a dynamic relationship between visitors and the exhibition. This is what we mean by an "exhibition that looks back”. Rather than a traditional passive viewing experience, the system engages in active conversation with visitors through a mobile web application using computer vision.

Visitors can simply take a photo anywhere in the exhibition and the system recognises both what they're looking at and where they are in the space. The large language model, trained on all exhibition content, offers context-sensitive information and reveals thematic connections across the 250+ projects. It tracks each visitor's interests, engagement times and decision patterns to tailor the experience in real time. For example, if you spend more time with environmental technologies, the system adjusts and recommends related works throughout the space.

These interactions create a living map of collective engagement. The data reveals which themes resonate most strongly, which projects generate unexpected connections and how different concepts link in visitors' minds. This transforms the exhibition into a learning organism that continuously deepens its understanding based on how it's being explored.

Jincy: How does the material palette for the 'nodal network' enhance the exhibition’s experience? Can you also delve into the flexibility of the modular column grid and how it supports both diverse exhibits and their integration with the Arsenale's architecture?

Niklas: The materials we worked with include recycled aluminium and 3D-printed wood composites. Different colours and material expressions are strategically deployed throughout the exhibition, creating distinct atmospheric zones that guide visitors through varying sensorial landscapes. For instance, our 3D-printed wood composites reference the briccole, the iconic mooring posts of Venice, connecting the exhibition's themes of circularity and renewal to the city's maritime heritage.

The column grid, based on a 50cm module, functions like a spatial operating system. It's flexible enough to accommodate over 250 diverse installations while visually aligning with the historic rhythms of the Corderie. This modular design approach embodies the exhibition's theme of Intelligens. Natural. Artificial. Collective. by demonstrating how flexible systems can adapt to changing requirements while maintaining structural coherence.

Jincy: What inspired the augmentation of Spatial Intelligens to transform individual experiences into a collective understanding of how we engage with architecture? How does the collective intelligence visualisation function within the space and what insights has it provided about 'architecture as a living, evolving system’?

Niklas: We wanted to make the exhibition aware of itself, functioning almost like a cybernetic organism. The inspiration came from recognising that individual curiosity, when aggregated, reveals collective patterns of engagement and generates new knowledge about how we understand architecture.

Technically, each visitor interaction is recorded, mapped and added to an evolving visualisation accessible through the app. This shows how the public navigates, dwells and connects with the works, transforming personal experience into collective insight. The system has revealed unexpected connections between projects that traditional curation might not highlight and shows which themes resonate most strongly across different audience demographics.

Spatial Intelligens functions as both mirror and memory for each visitor… Our aim was to create an exhibition that maps and responds to the complexity of thinking, both individual and shared.

Through this process, we demonstrate that architecture can behave like intelligent systems, not as a static container, but as a reflexive medium that adapts and learns. The visualisation acts as a feedback loop, helping both us and visitors understand architecture as something perpetually in formation through collective attention and engagement.

Jincy: Could you also comment on your collaboration with curator Carlo Ratti and how it influenced your design?

Niklas: Carlo and his team gave us a framework centred on intelligence, circularity and openness. The exhibition needed to accommodate an extraordinary diversity of thought and scale. We imagined the project as a living dataset and our task became one of spatial organisation: creating a modular, adaptable system that could structure a record number of participants without losing coherence.

Carlo's curatorial approach and his idea of a ‘fractal’ layout challenged us to think in terms of gradients of meaning. This aligned perfectly with our concept of creating both a physical network of nodes and a digital cognitive layer that could reveal the hidden connections between projects across the exhibition.

Jincy: How does this personalised, reflective element contribute to the larger proffering of architecture's interaction with 'natural, artificial and collective intelligence'?

Niklas: Spatial Intelligens functions as both mirror and memory for each visitor. It reveals what they gravitated toward, what they overlooked and how they moved through space, creating a unique map of their intellectual journey. Visitors discover, for instance, that they unconsciously focused on projects dealing with water management, or that they spent significantly more time in collaborative spaces than individual contemplation areas.

This reflective element transforms the exhibition from a one-time experience into an ongoing dialogue between body, mind and system. It embodies our larger narrative that architecture is not static, but a living medium continuously shaped by how we choose to engage with it, whether through natural curiosity, artificial augmentation or collective intelligence. The system creates a new form of spatial intelligence that reveals how we collectively understand and respond to our built world.

Architecture is not static, but a living medium continuously shaped by how we choose to engage with it, whether through natural curiosity, artificial augmentation or collective intelligence.

Spatial Intelligens posits intelligence—natural, artificial, collective—as subject and as operational, implicative infrastructure, where it’s no longer about just what’s shown, but how we show up. Architecture becomes a reflexive medium, capable of registering attention and recalibrating presence. It is a major pit-stop on the long-winded road to reconfigure the model of the architectural exhibition itself, how it is epistemologically constructed and experientially encountered, as an adaptive, interactive, co-constituted system; less a static display than a sentient interface. Conventionally, these have assumed a unidirectional flow: curator to content, content to viewer. This AI-driven intervention mutates that, with the Biennale and its performers executing a feedback loop of attention, activity and algorithmic inference. This is curation as computation, scenography as interface. What does it mean for an architecture biennale of this nature and scale to observe us as much as we observe it? What does it mean for it to accumulate our memories and learn our behaviours over time?

The 19th International Architecture Exhibition of La Biennale di Venezia is open to the public from May 10 to November 23, 2025. Follow STIR’s coverage of Venice Architecture Biennale 2025 (Intelligens. Natural. Artificial. Collective) as we traverse the most radical pavilions and projects at this year’s showcase in Venice.