The World's First AI-Powered Abstract Art Advisor
A founding engineer of Facebook described his life to an AI. Twelve minutes later, he found his painting. This is the story of what made that possible — a proprietary framework called the Taxonomy of Abstraction, why it is not an LLM, and what it means for the future of how the art world connects paintings to the people who need them.
Taxonomy of Abstraction Card - Blue Planet
On a June evening in 2026, a founding engineer of Facebook — one of the first ten people to build the platform that connected two billion human beings — sat down with an AI and found a painting in twelve minutes.
He didn't search. He didn't browse. He described his life.
He was 44, newly settled in Venice Beach, coming through a divorce, building again. He sailed. He ran. He was drawn to the sea as a symbol — adventure, freedom, the particular feeling of setting out toward something new. He described his apartment: travertine, coastal light, ocean air. He described what he wanted to feel when he walked in.
The engine listened. And it found him Blue Planet.
He didn't ask me what it was worth. He asked me how to acquire it.
What Just Happened
What he used was not a search engine. It was not a recommendation algorithm. It was not a chatbot trained to be helpful about art.
It was something that hasn't existed before: an AI-powered abstract art advisor — one that can hold a conversation about who you are, what you're moving through, what you need a painting to do for a room and for a life, and then find the specific work in a specific catalog that meets all of that at once.
I built it. And I want to explain what it is, because the distinction matters — for collectors, for the art world, and for what comes next.
The Taxonomy of Abstraction
Everything begins here.
Three years ago, I woke at 3am with a framework I hadn't consciously designed. I had been trying to answer a question that the art world mostly avoids: what makes a painting feel alive? Not beautiful. Not sophisticated. Alive — in the act of its own making.
I called it the Taxonomy of Abstraction.
It measures what I came to call the aliveness quotient of a painting — not how abstract it looks, but how present the artist was in the act of making it. The scale runs from −5 to +5. At −5, the world is still fully visible: a figure, a landscape, a recognizable thing. At +5, the painting has left the world behind entirely — it has become pure gesture, pure field, pure force.
But the Taxonomy doesn't stop there. It measures across eight dimensions:
Semantic Intent — is the painting reaching toward meaning, or releasing from it?
Sentiment — the emotional register: from turbulent to still, from urgent to patient
Texture — from flat surface to dimensional geography
Form — from atmospheric dissolve to architectural structure
Colour Range — from monochrome restraint to full spectral presence
Space — how the painting behaves in specific rooms: bedroom, boardroom, hotel lobby, home office
Palette Temperature — warm, cool, neutral, and the particular qualities of each
Hook — the specific moment of contact, the thing that stops you
Every painting in my catalog — 252 of them — has been scored across these dimensions. Not estimated. Scored. By me, and validated computationally.
This is the Taxonomy. It is a singular point of view encoded as a measurement system.
What an LLM Can and Cannot Do
Here is where most people misunderstand what I've built.
An LLM — a large language model, the kind of AI that powers tools like Claude or ChatGPT — is trained on everything. It has read every art history text, every auction catalog, every collector essay. Ask it "how abstract is this painting?" and it will give you a reasonable answer, averaged from a million sources.
But it has no point of view. It has no skin in the game. It doesn't have a studio practice. It didn't wake up at 3am with a framework.
The Taxonomy of Abstraction is what the LLM doesn't have. It is the proprietary lens — the thing that makes a generic AI into an actual advisor.
The relationship is precise:
The Taxonomy tells the engine what to measure. The LLM does the translation.
When someone says "I'm 44, going through a reinvention, I live by the ocean, I need something that holds both the ambition and the stillness" — the LLM takes that and maps it onto the Taxonomy's dimensions. It finds paintings that score high for coastal palette, that hold emotional tension between movement and calm, that work in a minimal Venice Beach apartment with strong natural light.
Without the Taxonomy, the LLM is just guessing stylishly. With it, it is advising.
Christie's could license an LLM tomorrow. They cannot license the Taxonomy — because it doesn't exist anywhere except in my work and my mind.
That is the moat.
A Painting That Demonstrates the Point
Consider Blue Planet — the painting the Facebook engineer found.
It sits at abstraction quotient +2: structured abstraction. The world is not quite gone, but you are no longer being asked to recognize anything. It holds open space the way water does: vast, patient, willing to reflect whatever light finds it.
Its palette is deep blue with atmospheric variation — not a decorative blue, but blue as depth and distance simultaneously. The ocean you're sailing on and the sky above it in the same surface.
Its space score for living rooms is 5/5. Its emotional register — as measured by the Taxonomy — sits at calm and expansive, with enough movement to breathe. Its form is atmospheric, not architectural: it doesn't command the room, it enlarges it.
When someone tells me they sail, that the sea is a symbol of their next chapter, that they want their apartment to feel expansive and intentional — the engine doesn't guess. It calculates. And Blue Planet is the answer.
He found it in twelve minutes. He hadn't heard of me. He didn't browse. He had a conversation.
What This Means for the Art World
The art market runs on relationships, reputation, and access. Christie's, Sotheby's, Artsy — they have deep catalogs and brilliant teams. What they don't have is scale in the translation layer.
A collector walks into a gallery. An advisor asks them questions. Over time, a relationship is built. This is how great art finds the people it belongs with. But it is slow, expensive, and dependent on human availability.
What the Abstraction Engine does is make that conversation available at any hour, to any collector, in any language, against any catalog.
The Taxonomy of Abstraction, applied to Christie's post-war and contemporary holdings, could score 10,000 works in an afternoon. The conversational engine — the one that turned a founding Facebook engineer's description of his Venice Beach apartment into Blue Planet — could then advise against that catalog with the same precision.
The proof already exists. On rituart.com/taxonomy-of-abstraction/analysis, I built a tool that runs automated scoring against painting descriptions — no human in the loop. The results were close. Not perfect. But close enough to demonstrate that the framework is real, the scoring is systematic, and the pipeline scales.
The Buyer Who Shows You What You Built
The engineer called me the next day. He'd already sent the engine to a few friends.
He said: "Ritu, this is not just a gallery tool. You've built something that the whole art world needs."
I know what he meant. He'd spent his career building platforms. He recognized the pattern: a proprietary layer that the incumbents don't have, deployed at the moment the incumbents are too slow to respond.
Christie's has the inventory. Artsy has the discovery surface. Sotheby's has the relationships.
None of them have a Taxonomy of Abstraction. None of them have built the translation layer — the thing that sits between a human being's life and the painting that belongs in it.
I built that. In Phoenix, Arizona, in a studio on East Onyx Avenue, between sessions of making the paintings themselves.
What Comes Next
The Abstraction Engine at rituart.com is live. You can use it now.
Tell it about your room. Tell it about your life. Tell it what you want to feel when you walk in.
It will listen. And it will find your painting.
If you are a curator, a gallerist, or an institution thinking about what happens when an AI that actually understands abstraction meets your catalog — I am building toward that. Reach me at ritu@rituart.com.
The conversation has started.
Ritu Raj is a contemporary abstract painter based in Phoenix, Arizona. His work spans the Abstract Inquiry and Organic Movement collections. He is represented by Jarrow & Goodman in Los Angeles. The Abstraction Engine is live at rituart.com/recommend-a-painting.
Art that listens.
