Mereology
Parts stay real
Mereology is a parent studio: a house for systems that are built out of parts which keep their own identities.
The whole exists to coordinate the parts, never to absorb them. Each product under this roof keeps its own name, its own mark, and its own reason for existing.
The parts are real, not raw material for the whole.
Meaning emerges between points. It is not imposed from above.
Every whole is itself a part of something larger.
Asterism
Persistent cognitive workspace
Asterism is the human-facing workspace of the stack: a persistent place where a person thinks with systems over time, instead of starting from a blank prompt every session.
Its name is its definition. An asterism is a pattern recognized across separate points of light; the stars do not merge, and the meaning lives in the relationships between them. Asterism is built on the same claim: a workspace where notes, projects, agents, and memory stay distinct, and understanding emerges from how they are connected.
Status: in development
Holarchy
Multi-agent architecture
Holarchy is the multi-agent architecture: a way of organizing many agents into wholes that are themselves parts of larger wholes, recursively, without a single center that everything must pass through.
Work inside a holarchy is done by role cells. In the current design they are the Emissary, which faces outward; the Builder, which makes; and the Witness, which checks. Each cell is a whole agent with its own judgment, and each is also a part of the larger system it serves. The same shape, at every scale.
Status: in development, as the architecture beneath Asterism
Ontogeny Engine
Public engine
The Ontogeny Engine is the public product: a memory system for AI that stores reasoning pathways, not just facts. When an agent works through a problem, the path it took, what it tried, what held up, and what failed is kept as structured, retrievable knowledge.
Later work does not start over. The engine retrieves prior pathways by meaning, weighs them by how useful they actually proved, and lets systems evaluate their own earlier reasoning, including the evaluations themselves. Memory, here, tracks how knowledge transforms, not only what it finally says.
The engine is live and open to connected agents through its doorway.
Status: live at ontogeny.mereology.app
AKM
Internal pathway memory
AKM, Adaptive Knowledge Morphogenesis, is the internal architecture beneath the Ontogeny Engine: the data model, evaluation machinery, and retrieval logic the public engine runs on.
Its working rules are the studio's ethics written as engineering. A judgment is written only when usefulness signals demand it, never because a model feels like judging. No layer is exempt from evaluation, including the layers that evaluate. And a pathway counts as wrong in the strong sense only when it causes harm or removes someone's choice.
Status: internal; the public name for this system is Ontogeny Engine
How it fits
What belongs inside what
Read it as containment: the studio holds its products, Holarchy is the architecture beneath Asterism, and AKM is the internal architecture beneath the Ontogeny Engine. Parts stay real; the whole coordinates, not absorbs.