Reflecting on Lineage

From Sufi mysticism to vector databases — finding lineage and wisdom in unexpected places

My grandfather wrote in 1968: "The problem with the hippies is they think they can take a magic pill and know more than their elders." My dad was a quintessential hippy — a star of the 1960s San Francisco psychedelic rock scene, so there was clearly some animosity in their relationship. Today, we might say the same about AI. Are we taking a digital magic pill, bypassing the hard-won wisdom of generations? Or is AI becoming something else entirely — a new vessel for transmitting lineage itself?

I was raised by my mom and step-dad who were deeply committed to their spiritual paths — my mom, a buddhist (following a Tibetan teacher), and my step-dad, a Sufi (following a western branch of Sufi mysticism) and an Indian classical singer (following an esoteric eastern music tradition). The word "lineage" was engrained into my upbringing — a word that, to me, means transmission and continuation, as well as respect and privilege.

"Everything we have now is the result of our ancestors who handed forth to us our language, the preservation of the land, our way of life, and our songs and dances."

— G. Peter Jemison, Cattaraugus Reservation Seneca Nation

I read Aldous Huxley's Island as a teenager, which influenced my view of society and how it could be different... possibly better. I was intrigued by the coming-of-age rituals portrayed in the book, in particular the climbing initiation, which nurtured a healthy relationship with fear. This planted the seed for a curiosity about coming-of-age rituals in different cultures. In some native cultures, people take three vision quests in their life: coming of age, transition to elder, and the final quest. The term "elder" in western culture typically invokes senior citizens, one step from the grave. In indigenous and spiritual traditions, elders are the holders of wisdom, the bridge between past and future generations, and cultural storytellers — revered members of the community.

This curiosity ultimately led to my taking a traditional Native American vision quest on my 21st birthday. The quest was deeply profound — I feel I could write a book about all my insights. A connection to lineage and elder knowledge was a core theme. My upbringing prepared me for this idea conceptually, but to actually feel the direct transmission was something altogether new. One of the immediate outcomes was feeling deep gratitude and respect for my parents — which was a revelation to my 21 year old self.

My next encounter with lineage came shortly after my vision quest as I followed my obsession with music. Having grown up in a family of musicians, I understood music lineage, and could map the evolutionary connections of many genres I loved. But the more I read about music, the more it converged into families of rhythmic and melodic ideas. Genres became generational stepping stones of musical lineage — embedded clues of each evolutionary step.

Mapping lineage became an obsession. First with music, then fonts and design theory, then programming languages. In the year 2000, while backpacking in Peru, I had a moment of inspiration where I connected my childhood library tool, the Dewey Decimal System, to my interest in building an experience that could explore contextual relationships. The Dewey system's elegant organization of knowledge into hierarchical, interconnected categories sparked an idea: what if we could navigate information not just categorically, but relationally?

Early concept of the Relation Engine discovery tool.

I called it the "Relation Engine" and built a prototype called Anaspace that would allow the user to navigate content relations through tree-like branchings. In this model, the physical position in the tree determined proximal connection to other nodes. This is a bit like going to the library, finding a topic you like, then turning around to discover a new topic that's related by its physical proximity, but possibly far from the prior subject.

This idea of relationship mapping and lineage has been a continuous theme in many projects over the years. Twenty years after that backpacking epiphany, AI finally gave me the tools to realize this vision. For my recent project SNDOUT, I revisited some of my early interest in mapping music lineage. This time using the embedded graph knowledge (connection vectors between entities) in AI models to represent a probabilistic view of the music lineage for bands and artists. For the first time I feel I have the tools to automate an idea that was always so daunting — mapping the lineage and connections between all things. It may be my favorite capability of generative AI.

Artist lineage map in my AI-powered music discovery tool SNDOUT.

Here's where the story takes a turn that even I didn't expect. Another recent project returns to my lineage roots, but this time the irony wasn't lost on me — using AI, the supposed "magic pill," to transmit authentic spiritual lineage.

I worked with my step-father to take a trove of writing from his teacher, Sufi mystic Murshid Samuel Lewis (known as Murshid Sam), and create a searchable, well organized archive of his life's work. The archive consists of 16,000 type-written pages of esoteric mystic philosophy, poetry, correspondence with global spiritual leaders and personal diary entries. He was recognized as both a Zen Roshi and Sufi Murshid — both meaning master teacher. His writing is both deeply profound and incredibly hard to understand without extensive initiation and conceptual grounding. He explains very little about what words or concepts mean — his writing is not intended to be a primer.

My step-father’s teacher Murshid Sam leading a group in dance.

A mix of curiosity and a desire to understand the deeper meaning led me to build an AI "guide" to translate the words and phrases into something clear and understandable. I put all of Murshid Sam's writing into a vector store — allowing the AI to draw connections across his entire corpus. I worked with the official archivist of his writing to refine the prompt and focus the tone and language to be authentic. We made a conscious decision not to anthropomorphize the guide — it's a stoic and helpful translator, nothing more.

The incredible value that emerges from this guide is the holistic and comprehensive map of all the ideas in his writing. While asking questions and receiving answers proved meaningful, it wasn't until I copied a sentence from his writing and pasted it in quotes after writing "please explain in the context of (paper name):" that I found the real value of this guide. I turned this into a widget that turn any selection into a contextual inquiry.

Selected copy invokes a widget that can send a contextual inquiry to the AI guide.

At one point I was exploring Murshid Sam’s thoughts on breathing, circadian rhythms and metabolic cycles, and I asked "how can I apply this to my practice of writing code?" The answer floored me. It gave me clear and practical steps to apply esoteric Sufi breathing practices to a cadence of coding that could give me more clarity, focus and energy. Things like synchronizing breath with code compilation cycles, breathing deeply after Claude Code botches a simple task, or working for 2-hour cycles to align with ultradian rhythms. As a curious skeptic, the jury is still out on the impact, but the direct transmission, interpreted into a modern context, from a teacher now deceased for 55 years, blew my mind.

You are welcome to explore the tool I built and the writings of Murshid Sam here: https://murshidsam.org. I'd love to hear what you think!

Personally it's one the most rewarding application of AI I've created (along with my SNDOUT project). I also feel this idea can apply to endless specialized domains where the general knowledge used to train LLMs fall short, or where specific context makes all the difference. But more than that, it made me reconsider what we're really building with AI.

What if AI is a new form of cultural memory? What if we're creating vessels that can hold and transmit the wisdom of our elders in ways that transcend the limitations of human memory and mortality? This raises profound questions for the foundation models: How should we think about training data as "ancestral knowledge"? What responsibility do we have to preserve not just information, but wisdom? And how do we ensure that AI amplifies genuine lineage rather than replacing it with synthetic substitutes?

Returning full circle, I went on my second vision quest this past May. My aim wasn't to claim elder status, but to revisit the connection to lineage and spirit. I was reminded that, even in this disembodied digital age of hyper-everything, we're just one small step away from a deep connection to our ancestors and ancient wisdom that's embedded in our DNA and the living world around us. Also remembering that disconnecting, slowing down, and sitting quietly, brings life and purpose into clear focus.

The view from my quest camp in the Inyo Mountains, east of the Sierra Nevada Mountains.

The paradox isn't lost on me: I spent four days alone in the wilderness seeking ancestral wisdom, only to return and build AI systems that encode and transmit that same wisdom. But maybe that's exactly the point.

Prompt: What would Murshid Sam say about AI preserving and transmitting the full context of human knowledge, complete with its spiritual lineages, relationships, and even its contradictions?

“Murshid Sam emphasized that spiritual knowledge is best transmitted heart-to-heart, soul-to-soul — a method that transcends mere words or traditional learning. Modern technology, like AI, may provide tools for sharing knowledge, but he would likely argue that it cannot replicate the spiritual essence that occurs in personal interactions.”

— Murshid Sam AI Guide (murshidsam.org)

Alan Watts said that we suffer from a hallucination of separateness, that we've forgotten we are the universe experiencing itself subjectively. Here's a short clip from his series The Essential Lectures Of Alan Watts. In this clip, Watts talks about the limits of language and math to understand the complexity of nature and the system we're all a part of. This reflects a parallel to the current state of generative AI, and the limits we can extract from language and math alone.

The question for those of us building with AI isn't whether it will replace human wisdom, but how we can design it to honor and amplify the lineages that brought us here. After all, models are trained on the echoes of our ancestors — their books, their code, their art, their conversations. Are we bypassing lineage or are we're encoding it? The challenge is doing so with the same reverence that any keeper of wisdom should hold.