nonstandardstudio

Speaking at ETH Zurich, Future of Construction Symposium 2026.

Most of the AI conversation in architecture is still focused on outputs.

I’ll be talking about something else:
what happens when workflows themselves become executable.

Showing this across research, teaching, and practice, and what starts to change once decisions are structured inside the system, not lost between steps.

More soon.

On March 13, 2026, I will be giving a guest lecture at the City Design Studio at Shanghai Jiao Tong University in Shanghai titled:

“Distributed Intent: Agentic AI and the Future of Architectural Reasoning.”

The lecture examines a conceptual shift currently emerging in the relationship between architecture and artificial intelligence. Rather than approaching AI primarily as a generative instrument, the discussion considers how design reasoning itself may become distributed across networks of specialized agents, knowledge systems, and evaluative processes.

Within this framework, architectural intent is no longer located exclusively in the individual designer or in a single model. Instead, it emerges through the orchestration of multiple interacting intelligences, each contributing different forms of analysis, generation, and evaluation.

The lecture will outline the theoretical foundations that inform the development of agentic design infrastructures such as Versur, and will explore how architects may begin to design not only spatial outcomes, but also the cognitive systems that produce them.

The discussion will be primarily theoretical and will focus on the conceptual models behind these emerging design systems.

Thank you to the City Design Studio team at Shanghai Jiao Tong University for the invitation. Looking forward to the discussion.

Hi friends,

While I’m still preparing Versur for public release, I’d like to share a closer look at how it works: a workflow-based, agentic AI creative engine designed to think, iterate, and build with you.

Currently in alpha testing, Versur continues to evolve as I refine how AI agents interact within creative workflows.

Chaining of AI Agents: Versur enables AI agents to be connected in sequences that mirror the nonlinear nature of design. Each agent is customizable: an Input Normalizer might structure your ideas, passing them to a Prompt Expander that explores variations, then onward to evaluators or visualizers that refine the results.

Together, these AI agents form a branching, iterative process that reflects how real creative work unfolds.

Nesting of AI Agents: Versur also allows AI agents to be nested within one another, coordinated by a general agent.

This creates a multi-layered system of collaboration: one that moves fluidly between global and local reasoning, abstract and specific intent, specialized and general perspectives.

It’s how Versur turns complexity into a coherent, adaptive creative environment.

These are just a few of the features shaping Versur’s agentic foundation: a system where multiple AI agents work in concert, not as separate tools, but as an evolving ecosystem that supports the full rhythm of creative thought.

Versur remains in its alpha phase, with a waitlist now open at https://versur.ai/ (check link in bio) for those interested in exploring what’s next for creative collaboration with AI.

I’d love to hear your thoughts: what kinds of creative workflows would you imagine building with AI agents? Share your ideas in the comments.

In architectural design, utilizing generic generative AI models presents unique challenges. Beyond the commonly recognized aesthetic biases/preferences these models might exhibit, a significant hurdle is generating multiple perspectives of a singular output while maintaining consistent composition.

Here, I'll highlight the capabilities of DeepHimmelblau, which can approximate various views of a single design.

This is a feature presented early this year (April) during the roundtable discussion on “AI and the Future of Design” https://youtu.be/jjUb48f4ROc?t=2883 with .patrik , , , , , and Wanyu He. Other features include coherent compositional relationships across interior, exterior, and elevations. To the best of my knowledge, this network was, and possibly still is, the only diffusion model capable of controlling views exclusively via prompts.

I've been working on the DeepHimmelblau project in collaboration with since 2017 with the aim of enhancing design processes and augment designers' creativity. DeepHimmelblau is a node-based system consisting of multiple neural networks tailored to various design tasks (Organizational – Technical – Gestalt).

I believe that as an industry, it’s imperative for us to develop our own tools, customized to our industry’s need, and geared towards pushing forward the performance of our industry. From tools dealing with design issues, to tools dealing with building performance, logistics etc.

https://onlinelibrary.wiley.com/doi/epdf/10.1002/ad.2808

The integration of AI into architecture is a game-changer, and I am confident in its potential to usher in groundbreaking changes. From Creative AI, creating intelligent design to optimizing building performance, AI promises to redefine the entire architectural process.

For those familiar with my research, I've been advocating for a long time for the development of industry-specific neural networks. Since 2017, I have been working on the DeepHimmelblau project in collaboration with Coop Himmelblau. Our objective is to enhance design processes and augment designers' creativity.

DeepHimmelblau is a node-based system consisting of multiple neural networks tailored to various design tasks (Organizational – Technical – Gestalt). The node-based structure enables task-dependent strategies of interconnected nodes to be established in response to discrete design tasks, specific design problems or the nature of any given investigation. The network not only allows the connection of multiple nodes, but also permits the semantic levels of the various node network layers to be combined, blended, and swapped between network nodes. One notable addition is the diffusion node, which was trained on CoopHimmelblau's proprietary image and text-based dataset. This node surpasses the limitations and aesthetic preferences of generic models, such as MJ, Stable Diffusion, and Dalle2.

Below is a collection of generated comparison outputs.

prompt:
"an interior circulation of a building made out of concrete, sinuous lines, aluminum triangular panelization, by CoopHimmelblau"

Here you can find more information about the project, and a roundtable discussion where I introduced some of the latest features of the DeepHimmelblau, such as establishment of coherent compositional relationships across interior, exterior, and elevations, ensuring compositional consistency and accurate approximation of different views of the same design.

https://onlinelibrary.wiley.com/doi/epdf/10.1002/ad.2808

https://youtu.be/jjUb48f4ROc?t=2883

Looking back on my Gaudi+NeuralNetworks, I can’t help but feel a genuine sense of amazement at the quality of the output that the generative adversarial networks (GANs) I developed achieved. While GANs are known to often produce mushy or flawed images, the networks developed for this project output results close to the quality of current diffusion models like , in terms of resolution and composition coherence.

It's great to see my research project Machine Perceptions: Gaudi + NeuralNetworks - one of the earliest exemplars of the application of and models in architectural design - receive the AIA Florida Merit Award for Theoretical and Research Professional! The Award recognizes outstanding contributors to the advancement of architectural theory and research, and I am honored to be this year’s recipient!

Biomorphic Series continued experiments.

Let's meet the speakers!
ACE Conference for

'With the advent of in , architectural design is ripe for . What would be the potential disruptive paradigm shift prompted by the use of and models in architectural design, and how would it manifest as unique -machine or -machine interaction protocols? Looking forward to get more insights from Daniel Bolojan, the founder of nonstandardstudio, Senior - Computational Design Specialist at CoopHimmelblau, an Assistant Professor of AI and Computational Design at Florida Atlantic University& #39;s School of Architecture, and a Ph.D. candidate at Die Angewandte Kunst in Vienna.

He is a leading voice in the implementation of AI strategies in architecture and the architectural design process. Over the years, he has taught several design studios and seminars at the Institute of Structure and Design-University of Innsbruck, Florida International University Miami, and has led numerous international workshops and conference workshops on the application of complex systems and Neural Networks in architectural design. He established his own research studio, Nonstandardstudio, in 2013. His design research has evolved over the years as a result of Nonstandardstudio& #39;s work at the intersection of generative design, computation, multi-agent systems, machine learning, and deep learning. In 2014, he joined the internationally renowned architectural practice CoopHimmelb(l)au as a Computational Designer, where he had the opportunity to work on numerous internationally renowned projects and competitions. Shortly after joining CoopHimmelblau, Daniel held the position of Junior Associate, Computational Design Specialist & Founder, and Head of Chbl|Code. As the head of Chbl|Code, he held the leading role of developing custom computational design tools, computational design strategies, virtual and augmented reality applications, machine learning tools and neural networks, and robotic fabrication processes. He is responsible for the office’s current drive to develop deep learning strategies aimed at augmenting the designer’s native abilities via the DeepHimmelblau neural network.

ACE Conference 'Architects for '
24 November 2022 - Brussels + online via ACE website
Discover the programme: https://bit.ly/3U74OmS
Register today: https://bit.ly/3TUFQXE
More info on ACE website.
With the support of Creative Europe

I made some great improvements to the network over the past week. Looking into the development of generative networks capable of learning relevant semantic features.

3D Domain Translation using Cycle-Consistent Adversarial Networks

Some of the algorithms I started developing while at I.sd Institute of Structure and Design, Innsbruck.

The 3D Domain Translation model starts to show some promising results. Although still more work is needed to be done to allow for a better disentanglement of features.

This approach builds on the work of Jun-Yan Zhu, Taesung Park, Philip Isola, Alexei A. Efros from Berkeley Ai Research (Bair) Laboratory, UC Berkeley

More detailed description soon...

@ Miami, Florida