Strategy 01
Work from your strengths
Use the capacities already available in the pattern and apply them with more skill and range.
AI SYSTEMS / SEMANTIC ASSESSMENT
Relic
I started by typing Adam and Eve.
A favorite work of Mark Twain, a journaling experiment, and vector embeddings became an AI-native Enneagram assessment—and the proving ground for a broader personality platform.
Launching August 2026
It began with a journal.
I wanted to build a journaling app that could do more than summarize what someone had written. You would describe a conflict, a decision, a relationship, or whatever occupied you, and the system would read the entry through the Enneagram.
The challenge was how to ground those insights in a coherent representation of the Enneagram instead of asking a language model to invent a reading on demand.
For the first experiment, I turned to Mark Twain’s The Diaries of Adam and Eve, long among my favorite pieces of literature.
Twain gives Adam and Eve unmistakably different voices. They notice different things, misunderstand each other differently, and want different things from the world.
I represented Enneagram patterns through vector embeddings, embedded the diaries, and compared each character’s writing with that reference system.
The readings were distinct and coherent. More importantly, the experiment suggested that vector space could do more than interpret journal entries.
It could provide a new foundation for the assessment itself.
That was when Relic became a product.
Scoring meaning, not answers.
Most personality tests use a hand-built point key.
Choose one response and three points go to one type. Choose another and two points go somewhere else.
Relic works differently. It asks:
Where does the meaning of this answer sit within the Enneagram?
I assembled and reviewed examples spanning all 27 instinctual subtypes—nine types across three instincts—and embedded them in a 3,072-dimensional meaning space.
Together, those vectors form a semantic map of the system.
Every possible assessment response is placed into the same space and measured against the full map. Two answers can use different words while expressing nearly the same motive. Two others can sound similar while pointing to very different patterns.
Relic scores those relationships rather than routing each answer into a predetermined bucket.
AI builds the instrument. The app runs it.
I did not want the assessment to send each answer to a language model for a fresh judgment. The scoring needed to be fast, exact, and repeatable.
So the semantic relationships are calculated in advance and compiled into score banks that run locally in the mobile app.
- 01Reviewed examples
- 02Vector embeddings
- 03Subtype geometry
- 04Compiled score banks
- 05Type, wing, and subtype
AI is used to construct a richer scoring instrument. The product then runs a fixed version of that instrument on-device.
This became a central engineering idea behind Relic: use AI where it adds expressive power, then compile that intelligence into something stable enough to become software.
A type should be useful.
Relic produces a type, wing, and instinctual subtype, then brings them together in a First Reading:
- 01What type are you?
- 02What this means
- 03What it implies your tendencies might be
- 04How to balance your type out
The result becomes the Relic itself, a visual object that preserves the shape of the reading and gives the user a place to return to it.
But the assessment is only the beginning.
Bring it a real situation.
A person can describe a conflict at work, a decision they keep circling, or a relationship in which the same difficulty keeps returning.
Relic reads the situation for the Enneagram patterns active within it, then compares that situational signal with the user’s established type, wing, and subtype.
It returns two genuinely different strategies.
Strategy 02
Move toward balance
Use explicit Enneagram movement to interrupt the reflex most likely to dominate the situation and open a different route.
The Enneagram is the proving ground.
Relic is the first complete implementation of a larger platform for making personality and assessment systems AI-native.
The architecture can take a framework, represent it in embedding space, compile it into a scoring system, combine a stable profile with new contextual input, and expose the result through an API or MCP.
The same approach could support Big Five, strengths frameworks, coaching systems, team models, or proprietary assessments.
Framework creators bring the model. The platform provides the embedding pipeline, scoring layer, profile structure, contextual readings, and agent interface.
Relic is the first end-to-end demonstration—from semantic representation to a working mobile experience.
Owen Fowler
AI Systems Builder
Tell me what you’re working on.
A few sentences is enough.