diff --git a/README.md b/README.md
index a8065a6c768..a504cba87eb 100644
--- a/README.md
+++ b/README.md
@@ -4,17 +4,25 @@
 ![TypeScript](https://img.shields.io/badge/TypeScript-5.x-blue.svg)
 [![Docs](https://img.shields.io/badge/docs-objectstack.ai-0a0a0a.svg)](https://objectstack.ai/docs)
 
-> ## Apps small enough for AI to hold whole.
+> ## The ontology is the software.
+>
+> **One executable business ontology.** AI writes it, the runtime runs it, agents
+> operate it, you own it.
+>
+> 本体即软件。一份可执行的业务本体。AI 写，运行时跑，Agent 用，归你所有。
 >
 > ObjectStack turns the whole app — data model, UI, workflows, permissions —
-> into typed metadata that fits in a single context window. Agents read it
-> whole, reason it whole, refactor it whole.
+> into typed metadata that fits in a single context window: apps small enough
+> for AI to hold whole. Agents read it whole, reason it whole, refactor it whole.
 >
 > That metadata is your **business ontology** — an open, versioned definition of
-> your objects, permissions, and flows that you own, not code scattered across a
-> framework. Strict TypeScript, Zod schemas, and a validation gate catch the
-> agent's mistakes at authoring time; the runtime derives the database, REST API,
-> UI, and MCP server, and enforces permissions and audit on every call.
+> your objects, relations, actions, permissions, flows, and agent and tool
+> definitions that you own, not code scattered across a framework. It is
+> executable, not a knowledge-representation ontology: no inheritance, no
+> axioms, no reasoner. Strict TypeScript, Zod schemas, and a validation gate
+> catch the agent's mistakes at authoring time; the runtime derives the
+> database, REST API, UI, and MCP server, and enforces permissions and audit on
+> every call.
 
 `Fits in an agent's context` · `Typed, validated, governed` · `Self-host anywhere` · Apache-2.0
 
@@ -152,7 +160,7 @@ The reason this works is the same reason TypeScript was the right host language:
 **an agent's errors become located, corrective text it can read and fix itself**,
 in seconds — instead of a silent runtime failure nobody traces back.
 
-The other half is size. The bundled example CRM — [`examples/app-crm`](./examples/app-crm):
+The other half is size: apps small enough for AI to hold whole. The bundled example CRM — [`examples/app-crm`](./examples/app-crm):
 objects, views, a dashboard, a lead-conversion flow, permission sets, actions,
 translations — is small enough for an agent to load end-to-end, reason about
 every dependency, and refactor across data, API, UI, and permissions in one
@@ -163,9 +171,18 @@ hoping. Measure it yourself:
 find examples/app-crm/src -name '*.ts' -not -name '*.test.ts' | xargs cat | wc -l
 ```
 
-> Your objects, permissions, and flows are your business ontology — and the
+> The ontology is the software. Your objects, relations, actions, permissions,
+> flows, and agent and tool definitions are your business ontology — and the
 > definition layer of the AI era should be an open protocol you own.
 > [Read why](https://www.objectos.ai/en/blog/ai-ontology-open-protocol/).
+>
+> What that does and does not mean: the ontology is *executable*, so it is
+> validated rather than reasoned over — there is no object inheritance, no
+> axioms and no reasoner. It is not a semantic layer over your existing systems
+> (federating an external datasource is read-only by default and early). Views,
+> dashboards, apps, and translations are projections of the ontology, not part
+> of it. And code does not disappear: it moves into the runtime — hooks, action
+> bodies, CEL, and constrained JSX.
 
 ## Your app is AI-operable, for free
 
diff --git a/apps/docs/app/[lang]/page.tsx b/apps/docs/app/[lang]/page.tsx
index 9121283d864..2c88bdcd05d 100644
--- a/apps/docs/app/[lang]/page.tsx
+++ b/apps/docs/app/[lang]/page.tsx
@@ -33,7 +33,7 @@ const mono = IBM_Plex_Mono({
 
 const HOME_TITLE = 'Metadata framework for AI-written apps';
 const HOME_DESCRIPTION =
-  'ObjectStack turns the whole app — data model, UI, workflows, permissions — into typed metadata that fits in a single context window.';
+  'One executable business ontology. AI writes it, the runtime runs it, agents operate it, you own it. The whole app — data model, UI, workflows, permissions — is typed metadata small enough for AI to hold whole.';
 
 /**
  * The homepage's metadata. Its social card is the shared hero cover,
@@ -212,15 +212,16 @@ export default function HomePage() {
             className="mx-auto mt-5 max-w-3xl text-[2.6rem]/[1.06] font-bold tracking-tight text-balance md:text-6xl/[1.04]"
             style={{ fontFamily: 'var(--l-display)' }}
           >
-            Apps small enough{' '}
+            The ontology{' '}
             <span className="block bg-gradient-to-r from-indigo-500 to-purple-500 bg-clip-text text-transparent dark:from-indigo-400 dark:to-purple-400">
-              for AI to hold whole.
+              is the software.
             </span>
           </h1>
           <p className="mx-auto mt-6 max-w-2xl text-lg text-fd-muted-foreground text-pretty">
-            ObjectStack turns the whole app — data model, UI, workflows, permissions — into
-            typed metadata that fits in a single context window. Agents read it whole, reason
-            it whole, refactor it whole.
+            One executable business ontology. AI writes it, the runtime runs it, agents operate
+            it, you own it. The whole app — data model, UI, workflows, permissions — is typed
+            metadata small enough for AI to hold whole: agents read it whole, reason it whole,
+            refactor it whole.
           </p>
           <div className="mt-8 flex flex-wrap items-center justify-center gap-3">
             <Link
@@ -294,10 +295,12 @@ export default function HomePage() {
         {/* ── what that metadata is, in prose ──────────────────── */}
         <section className="relative mx-auto w-full max-w-5xl px-6 pt-12 md:pt-14">
           <p className="text-base text-fd-muted-foreground text-pretty">
-            That metadata is your business ontology — an open, versioned definition you own.
-            Strict TypeScript, Zod schemas, and a validation gate catch the agent&apos;s mistakes
-            at authoring time, and the runtime derives the database, REST API, UI, and MCP
-            server — permissions and audit enforced on every call.
+            That metadata is your business ontology — an open, versioned definition of your
+            objects, relations, actions, permissions, flows, and agent and tool definitions that
+            you own. It is executable, not a knowledge-representation ontology: no inheritance,
+            no axioms, no reasoner. Strict TypeScript, Zod schemas, and a validation gate catch
+            the agent&apos;s mistakes at authoring time, and the runtime derives the database,
+            REST API, UI, and MCP server — permissions and audit enforced on every call.
           </p>
         </section>
 
@@ -447,8 +450,10 @@ export default function HomePage() {
         <section className="relative mx-auto w-full max-w-6xl px-6 pt-8 pb-20 md:pb-28">
           <div className="rounded-xl border border-fd-border bg-fd-card/60 px-6 py-8 text-center md:px-10">
             <p className="mx-auto max-w-2xl text-sm text-fd-muted-foreground text-pretty">
-              Your objects, permissions, and flows are your business ontology — the definition
-              layer of the AI era should be an open protocol you own.{' '}
+              The ontology is the software. Your objects, relations, actions, permissions, and
+              flows are your business ontology — views, dashboards, apps, and translations are
+              projections of it, and the definition layer of the AI era should be an open
+              protocol you own.{' '}
               <a
                 href="https://www.objectos.ai/en/blog/ai-ontology-open-protocol/"
                 className="inline-flex items-center gap-1 font-medium text-fd-foreground underline underline-offset-4 transition-colors hover:text-fd-primary"
diff --git a/content/docs/concepts/index.mdx b/content/docs/concepts/index.mdx
index c0fa1f42ccb..19b7e11d0f5 100644
--- a/content/docs/concepts/index.mdx
+++ b/content/docs/concepts/index.mdx
@@ -94,6 +94,11 @@ From this single definition, ObjectStack automatically:
 2. **The API is a Consequence** — REST endpoints appear automatically from object definitions
 3. **The Schema is the Application** — Your entire business logic lives in metadata files
 
+In the enterprise-AI vocabulary, the ontology is the software: the core (objects,
+relations, actions, permissions, flows, agent and tool definitions) is what the runtime
+executes, and views, dashboards, apps, and translations are projections of it. See
+[Business Ontology](/docs/getting-started/glossary#business-ontology).
+
 ### When to Use Metadata-Driven
 
 ✅ **Great for:** CRUD apps, SaaS platforms, admin panels, rapid prototyping, multi-tenant systems
diff --git a/content/docs/concepts/metadata-driven.mdx b/content/docs/concepts/metadata-driven.mdx
index 4e59d6c35d9..c0a7d1a6d01 100644
--- a/content/docs/concepts/metadata-driven.mdx
+++ b/content/docs/concepts/metadata-driven.mdx
@@ -151,6 +151,14 @@ Your entire business logic lives in:
 
 The Kernel simply **interprets** these definitions.
 
+In the enterprise-AI vocabulary, the ontology is the software. The ontology core is
+objects, relations, actions, permissions, flows, and agent and tool definitions; views,
+dashboards, apps, and translations are projections of it, not ontology. The ontology is
+executable, so it is validated rather than reasoned over — there is no inheritance, no
+axioms and no reasoner — and code does not disappear, it moves into the runtime (hooks,
+action bodies, CEL, constrained JSX). See
+[Business Ontology](/docs/getting-started/glossary#business-ontology).
+
 ### 4. The Tool Surface is Generated
 
 **Traditional:** Every AI tool is manually described and wired to custom queries  
diff --git a/content/docs/getting-started/glossary.mdx b/content/docs/getting-started/glossary.mdx
index fb14b9f1930..36397da00ef 100644
--- a/content/docs/getting-started/glossary.mdx
+++ b/content/docs/getting-started/glossary.mdx
@@ -91,8 +91,20 @@ arrived from that vocabulary, this is the term's referent here: the **Source** y
 which the runtime derives the database schema, REST API, UI, and MCP server. See
 [Metadata-Driven Development](/docs/concepts/metadata-driven).
 
-The word is worth stating explicitly because ObjectStack's differs from the hosted
-ontologies the category otherwise means by it, on two points:
+The slogan this reduces to is **the ontology is the software**, and the word is worth
+stating explicitly because ObjectStack's differs from what the term otherwise means, in
+two directions.
+
+**Against a knowledge ontology** (OWL, RDF, a knowledge graph): a knowledge ontology
+*describes* a domain, is read-only to the systems that consult it, and is *reasoned over*
+by an inference engine. An executable ontology *defines* the app, is *run* by the
+runtime, and is *validated* rather than reasoned. ObjectStack has no object inheritance,
+no axioms and no reasoner (the `abstract` key was removed from the spec for exactly that
+reason, ADR-0049), and it is not a semantic layer over your existing systems —
+federating an external datasource is read-only by default and early (see
+[External Datasources](/docs/data-modeling/external-datasources)).
+
+**Against a hosted ontology** (a graph held inside a vendor's system), on two points:
 
 * **You own the definition.** It lives in your repository as ordinary TypeScript,
   versioned in your VCS and reviewable as a diff, in a format specified and licensed
@@ -104,6 +116,12 @@ ontologies the category otherwise means by it, on two points:
   of behaviour implemented somewhere else, so it cannot drift from the system it
   describes.
 
+**Core and projections.** The ontology core is objects, relations, actions,
+permissions, flows, and agent and tool definitions. Views, dashboards, apps, and
+translations are *projections* of that core — authored in metadata too, but not
+ontology. And code does not disappear: it moves into the runtime, as hooks, action
+bodies, CEL expressions, and constrained JSX.
+
 Distinct from **semantic layer**, which in these docs means the narrower analytics
 `dataset` layer (ADR-0021) that reports and dashboards bind to — see
 [Analytics & Datasets](/docs/data-modeling/analytics).