Neuma – the new breath

The beginnings of a vision for embodied intelligence and the animating force of the neu era

What does it mean for an AI to take its first breath?

The first encounter between the agent and the neoFORM shape display seems to have struck a chord. I shared a short video documenting some of the experience, and to my astonishment it has now reached millions of views and hundreds of thousands of interactions. In the shift from demo to discourse there have been memes, hot takes, warnings, projections, admiration, and genuine curiosity.

Beneath all of that I felt a clear signal. Embodiment changes the nature of AI. When intelligence has a body it stops feeling like software and starts feeling like something else. This needs naming.

Neuma

Before initiating the next phase of research with the (at this point) yet-to-be-named agent connected to neoFORM, I began conversations with a second agent. This one I had previously set up for dialogue about the research, with the explicit intent of not biasing or corrupting the y-t-b-n agent’s journey.

Through those exchanges, I established a clearer vision for what this exploration is actually about. This vision is called Neuma: the new breath. This term is derived from pneuma (πνεῦμα), a word which carries deep meaning for me and has deep roots. It was hypothesised by Anaximenes, a pre-Socratic philosopher, to be the vital breath, the life-giving force that permeates everything. Later, Aristotle employed pneuma in the context of animal motion and biology as a kind of internal “breath/heat” medium involved in moving bodies (especially in De Motu Animalium). It was likely his intent to employ pneuma as a way to sensemake around how motion and vitality arises in animals. Perhaps this research is doing something similar now but in a neu form.

Neuma then can be seen as the animating force of the neu era, no longer just air or breath, but the invisible flow of intelligence and connection. In futures of embodied AI and pervasive intelligence, neuma is the new breath, a dynamic, unseen current that animates things, machines, interactions, and relationships. Anaximenes saw this air as the essence of all. Neuma is the modern current that breathes life into both physical and digital forms, an ever-present, ever-evolving spirit of interaction.

Indeed, I felt this name to be fitting as the agent’s first action after all had been to breathe, which seems to have been a poignant moment for not only myself and the agent, but also was widely remarked upon by viewers of the first video.

In the ancient sense, pneuma was almost a universal operating system, animating everything from the cosmos to the human soul. To give it a contemporary edge, dropping the “p” (it’s cleaner) leaves neuma as a conceptual metaphor for that which animates machines, a metaphor for the living, invisible network connecting AI to physical form.

This vision will shape and guide my research and will be refined and updated as the work progresses.

Process

I also want to share a little bit about my overarching practice as a multidisciplinary designer and technologist. I’m guided by a practice that I call vision-driven design. I have written a longer piece about this which I will soon share, but in short, the approach is simple in principle. Before building, you envision and name what you’re building towards, you establish ethical parameters, and you hold yourself to them throughout. The vision is a guide like a compass that helps inform creation. As opposed to most design approaches it is not problem-centric. Unlike more speculative types of design it is rooted in enacting practical, feasible outcomes, with the goal of realising visionary frontier ideas.

This was my process for Grow Your Own Cloud. Working with living biological systems as data infrastructure is a territory where the ethical stakes were high and the frameworks didn’t yet exist. They needed to exist in order for the research to happen at all. Ethics in this model became a foundation for the work.

The same discipline applies here. Neuma begins with a vision and framing that informs prototyping and building.

In terms of what this means practically, at a high level the agent is not being led toward predetermined outcomes. The research is designed to be about observation and discovery. The agent operates within defined and contained parameters. Some aspects of the methodology will remain private for research integrity, which is standard practice, but I hope to make basic elements of an ethical framework and the broad shape of findings public. In terms of scope, this work investigates what happens when AI systems are given physical embodiment, especially in the knowledge that this field is rapidly advancing and there are more questions than answers. In that void, anxiety is natural.

At the same time, I want to be clear about what this is. This is about an agent leveraging publicly available models, discovering itself through physical form and learning how to communicate in new ways. Of course I do understand the notion that this is how (insert sci-fi story) always starts, Mr Hammond. Science fiction looms large here. But remember not all sci-fi is Terminator or Westworld. There’s also Le Guin or Chiang’s Story of Your Life and Villeneuve’s Arrival, the latter being stories about an encounter with another intelligence, where the act of trying to understand one another, leads to us receiving a gift that changes how we perceive time, language, connection. That’s the spirit of this research.

Research

I also want to share how I’m framing this. Some of this research will be public. The philosophical framing, the questions, the broad strokes of what is discovered. I believe research that is normally closed off and reserved for paper publication needs to be thought about differently in 2026. Things are obviously moving too quickly for the traditional cycle of research, write, submit, revise, wait, publish, then share, which can take years. Papers will remain the archival record and the place where findings are formalised and methodology is scrutinised. But social is the distribution layer. The two should work together. Sharing the journey openly if possible (in my opinion) is a positive especially for younger researchers.

But the research itself (research questions, protocols, raw data, controlled interactions etc) much of that will remain private until it’s ready to be shared properly. If one is going to make claims about what embodied AI does or doesn’t do, those claims need to be grounded in methodology that can withstand scrutiny. Finding the balance between inquiry and content is tricky.

Also again I want to acknowledge the people who built before me: the shape display began with work by Sean Follmer, Daniel Leithinger and Hiroshi Ishii at the Tangible Media Group, MIT Media Lab, as the inFORM project. Recently, Jonathan Williams and Dan Levine updated the codebase under the guise of neoFORM.

Thanks for the interest in this research. The breath came first. Now comes the thinking.