Semantic Types

Things, their kinds,
and how they relate.

Semantic Types models the world as facts - a human-readable model that runs live, ready for programs and AI agents to build upon.

the lab behind x10sibility - for software engineers, architects and data modelers

Explore the idea Read the essays
Things The actual ones - this car, that invoice, your plant in Hamilton. Kinds Types at any depth, each narrowing the one above, inheriting everything. relations The expressive ways they connect - far beyond simple composition.

01 · Things & kinds

A specific thing narrows its kind.

It inherits everything and states only what differs. MyRAV4 is a Toyota, so it has a color - and it can only be one of Toyota's colors. Types and instances are deliberately blurred: a specific thing is just a further narrowing.

primaryassociationtarget CarhasOneColor ToyotaCarisACar ToyotaCarhasOneToyotaColor MyRAV4isAnInstanceOfToyotaCar MyRAV4hasOneSupersonicRed
ThingsMyRAV4, SupersonicRed KindsCar, ToyotaCar, Color, ToyotaColor RelationshasOne, isA, isAnInstanceOf

Everything is a fact: primary, association, target. Facts can arrive in any order, from any source - the result is the same.

02 · How they relate

Beyond composition.

Most systems are built from one relation - composition. Semantic Types relates things in more expressive ways, and relations are data too: a new one specializes an existing one and behaves like it.

Read Relationships →
isAa kind of
isAnInstanceOfan actual one of
has · hasManyis made of
hasOwnowns, for life
containsholds, for now
hasOneRefrefers to
requiresmust have

03 · From the lab

Built here. Run by x10sibility.

Semantic Types is the lab: the research, the engine and the modelling spec. It isn't sold on its own. x10sibility runs it and opens it to customization.

runs · for builders x10sibility Operations today - hosting, sign-in, backups. Next, the marketplace and citizen-developer portal.

Principles

Everything is an entity

Types, instances, relations, rules, users. The richness comes from how they are associated.

Narrowing, not copying

Say it once at the kind. Everything below inherits; exceptions are stated where they occur.

Build on what exists

Existing databases are mapped and enriched in place, not copied. Models build on other models.

AI at authoring time

Models help write mappings and rules; a person approves them; only approved rules run. Same input, same result.

Writing

Why we build lists Clay tablets, the periodic table, parts lists - and why a list with details quietly becomes a model. Relationships From Aristotle's "is" and "has" to the relations a model needs, and the patterns built on them.
All articles →