Technology and Healing have often combined
My career pathway has been somewhat circuitous and has followed many different threads. Technology and experiential, embodied practice have been intertwined in my life since the early '90s, when I started making electronic music.
At various times in my career, the threads of technology and healing have woven in and out of each other, but they have never really been far apart. Some of those threads include working as an acupuncturist in the early 2000s while creating CD-ROMs and audio CDs to help students learn the names of acupuncture points. At other times I have been a full-time web developer, employing people across the globe to create templates for a content management system whose brand revolved around nature, eco-friendly imagery, and minimal design. I have worked as a UX/UI gun for hire while also running Holotropic Breathwork workshops on the weekends, and I now work full-time as a psychotherapist, run Holotropic Breathwork workshops every two months in Melbourne, Australia, and continue to tinker with a few different web- and app-based creative projects.
Almost a decade of the Music for Breathwork website
Technology is, of course, at the core of the Music for Breathwork website. It's approaching ten years of operation, and over that time the technology behind it has evolved enormously.
One thing that hasn't really changed, however, is the importance of curation. Every track in the catalogue is listened to by a human being - mainly me. Every description and every tag is written by me. I experimented with AI-generated descriptions and tagging early on, but something always felt off. The descriptions tended to become generic, and the tags often missed the qualities that matter when you're selecting music for an expanded state of consciousness. There are still tracks in the catalogue that were generated during those experiments, and I'm gradually working my way through them, rewriting descriptions and correcting tags by hand. Behind the scenes, a significant amount of the work involved in maintaining the catalogue isn't simply adding new music - it's continually refining what is already there.
Software development has changed over the last 20 years
Where AI has genuinely transformed my work is in software development. When I first started building websites over twenty years ago, almost everything was hand coded. Over time that changed with frameworks, libraries, and better development tools, and now AI has become another step in that evolution. I still need to understand how the code works, how the pieces fit together, and whether the solution is appropriate, but AI allows me to build things that simply wouldn't have been possible on my own.
The Limina apps are a good example. Without AI they probably still would have been built, but it would likely have taken years rather than weeks, and I would almost certainly have needed to employ other developers to help with the more technically demanding parts. For me, AI is a means to an end. It allows me to spend less time writing repetitive code and more time thinking about the problems I'm trying to solve.
Editing and proofing text is invaluable
The same applies to writing. I use AI to proofread, edit and tighten course material, blog posts and marketing copy. It helps improve clarity, catch mistakes and smooth awkward sentences, but it operates within very tight boundaries. I have little interest in publishing content that wasn't genuinely written or conceived by me. The ideas, the experience and the perspective still need to come from somewhere.
There is no AI in the generation or curation of the catalogue or playlists
The one place where I deliberately avoid AI is the musical curation itself. Every track begins with listening. I spend time understanding the emotional arc of the music, selecting the key descriptive tags, and then using a weighted relationship map that I built to suggest other relevant tags. Those suggestions aren't generated by an AI model - they come from relationships that I have defined over many years of working with music in therapeutic settings. The suggested tags are then reviewed, adjusted and often rejected before the track is finally saved.
The screenshot above details the tag relationships for the "Dark" tag. In order for a track to be suggested to be dark it will always have the mysterious tag, or needs both introspective and somber and so on.
The automatic playlist generation works in much the same way. Although it might appear intelligent from the outside, it isn't AI. It is a system built on thousands of hours of human listening, careful tagging, mathematical relationships between audio features, and an understanding of how music unfolds in practice. Sometimes the results are surprising, and sometimes they miss the mark, but the goal has never been to create a perfect machine. The goal has always been to build tools that extend human judgement rather than replace it.
There should be no AI generated music in the catalogue
A general principle I stand by is that there should be no AI-generated music in the catalogue, and to the best of my knowledge this is indeed the case. I firmly believe that I want AI to help with the laundry rather than make my art. So, in principle, even if an AI-generated track met all the requirements of a Holotropic Breathwork piece, I would not allow it to be added to the catalogue.
If you come across anything that looks like it may be AI-generated, please let me know and I will investigate it as soon as I can.
Who knows what the future will hold
Technology will continue to change. The tools I use to build the website, develop software, write documentation or edit my own work will almost certainly look very different again in another ten years. I'm happy to embrace those changes where they genuinely make the work better.
What I don't want to automate is the part that gives Music for Breathwork its value and for me the most joy. Listening to music, understanding its emotional landscape, recognising where it belongs in a breathwork session, and describing it in a way that helps another facilitator or breather find exactly what they need - those are still profoundly human tasks.
AI has become an invaluable part of my workflow, but it isn't the thing that curates the catalogue. It isn't the thing that understands the music. It isn't the thing making creative decisions. Those decisions still come from thousands of hours of listening, facilitating, making mistakes, refining the system and, above all, remaining curious about the relationship between music and human experience.
For me, that's exactly where technology belongs: not replacing human judgement, but supporting it.