One ontology that classifies any enterprise data. Any system. Any industry. No custom data models. No exceptions.
6
universal primitives
40
disjoint axioms
194
taxonomy concepts
100%
classification accuracy
SCROLL
The Problem
Canonical data models never hold
Every enterprise eventually tries to build one unified data model. It takes years, costs millions, and is obsolete before it launches. LEAPED takes a different position: don't model your data, classify it.
Traditional approach
6–18 months to define a canonical model
€2–5M consulting before a single record maps
Model is obsolete before it launches
Every new source requires manual mapping
AI can't reason over inconsistent schemas
LEAPED approach
Six primitives classify any record instantly
Ontology-driven — the world model the LLM reasons through
100% classification accuracy on real data
New sources auto-classify via data signals
AI reasons natively over the knowledge graph
The Foundation
Six primitives. Mutually exclusive. Collectively exhaustive.
Every record in every system in every enterprise maps to exactly one. No exceptions. Proven by 40 disjointness axioms in OWL. 100% classification accuracy on production data.
📍
Location
The stage. Never changes. Provides context for everything else.
canBeChangedByEvents: false
👤
Entity
The actor. Only primitive that can trigger events. Agency.
canTriggerEvents: true
💎
Asset
The instance. Acted upon, never acts. Patient, not agent.
instanceOf → Product
📦
Product
The blueprint. Specification, template, type definition.
canBeChangedByEvents: false
⚡
Event
The moment. Where all primitives intersect. Sole causation.
THE HYPEREDGE
📄
Document
The evidence. Records, proves, communicates. Inert without events.
requiresEventToAct: true
The Breakthrough
⚡
👤
Entity triggers
💎
Asset changes
📄
Document records
📦
Product concerns
📍
Location context
🏢
Location occurs at
The Hyperedge — collapsed reality
An Event isn't a kind of thing. It's the moment where all other primitives intersect. Who triggered it, what changed, where, which product, documented by what. One node that connects everything. This is how enterprises actually work — across every system, every department, every silo.
17 Golden Types
What it IS — intrinsic, exactly one
Golden Type determines what a record fundamentally is. Golden Role determines what function it serves. One type, multiple roles.
Location
🏢Physical
☁️Digital
Entity
🧑Human
🏛️Enterprise
🏬Organization
⚙️System
🤖Agent
Asset
🔧Physical
💻Digital
💰Financial
Product
📦Good
🛠️Service
Event
⚡Atomic — 7 base subtypes + domain
Document
🧾Transactional
📝Contractual
🚢Operational
📋Informational
🎬Media
How It Works
From raw data to knowledge graph
Data signals in the field names tell LEAPED what it's looking at. Classification happens at ingestion — no design phases, no schema debates, no manual mapping.
01
Raw Data
Any field from any system
→
02
Data Signals
Pattern match field names
→
03
Golden Type
What it IS
→
04
Golden Role
What it SERVES
→
05
Knowledge Graph
Queryable by AI
Scene 01 — Logistics
Cold chain: order to delivery
A freight order flows from CargoWise through Boltrics to delivery. Three systems. One knowledge graph. Every temperature excursion traceable.
⚡
CreationEvent — Freight order created
Customer places order in CargoWise. The shipment lifecycle begins.
🏛️ Seafood Imports BV📄 Freight Order #4821📦 Frozen Shrimp 40ft📍 Port of Rotterdam
⚡
MovementEvent — Container loaded
Reefer container loaded at origin. Temperature monitoring starts. Asset in transit.
🧊 Container MSKU-4471🏛️ Maersk Line📍 Port of Shanghai → Rotterdam📄 Bill of Lading
⚡
ObservationEvent — Temperature excursion
Sensor logs -14°C. Threshold is -18°C. Breach detected automatically by ontology rules.
🧊 MSKU-4471 → BREACH⚙️ IoT Sensor #892📄 Temp Log
⚡
TransactionEvent — Goods received in Boltrics
Container arrives at warehouse. Boltrics WMS confirms receipt. Cross-system: same Asset, same graph.
A lead in HubSpot becomes a customer in Salesforce becomes a debtor in the ERP. Three systems call it three things. LEAPED calls it one Entity with roles.
⚡
CreationEvent — Lead captured in HubSpot
New contact created. HubSpot calls it a "Contact". LEAPED classifies: Entity.Human, Role: Contact.
👤 Jane van Dijk📄 Form submission☁️ HubSpot CRM
⚡
UpdateEvent — Qualified in Salesforce
Same person, different system. Salesforce calls it an "Account". LEAPED: same Entity, new Role: Customer.
👤 Jane van Dijk → Role: Customer📄 Opportunity #SF-2847☁️ Salesforce
⚡
ContractEvent — Deal signed
Contract executed. Document links Entity, Product, and Event. One node connects everything.
📝 SaaS Agreement 12mo📦 Platform License Enterprise👤 Jane van Dijk💰 License #LIC-0091
⚡
UpdateEvent — Customer in ERP
ERP calls it a "Debtor". LEAPED: still the same Entity. Three systems, one knowledge graph — a lead became a customer.
👤 Jane van Dijk → Role: Customer🧾 Customer Record #CUS-0341💰 Account €48,000☁️ NetSuite
Enterprise domain agnostic
Same six primitives. Any enterprise. Any industry.
No custom data models. Every enterprise runs on the same six primitives — only the vocabulary changes.
🚢
Logistics & Supply Chain
Carrier = Entity · Container = Asset · Shipment departure = Event · Port = Location · Bill of Lading = Document · 40ft Reefer = Product
LEAPED isn't just software. It's a shared language for enterprise architecture — usable in workshops, strategy sessions, and implementation alike.
🃏
The LEAPED Card Game
Six suits. Six primitives. Teams classify real business scenarios into Location, Entity, Asset, Product, Event, and Document. Builds shared understanding in 60 minutes — no technical background needed.
workshop tool
🏗️
Architecture Workshops
Map your enterprise landscape using LEAPED primitives. Identify data silos, trace value streams, and design integration strategies — all in the same six-word vocabulary.
strategy session
🔄
From Workshop to Platform
Everything mapped on the whiteboard translates directly into the knowledge graph. No lost-in-translation. The workshop output IS the implementation input.
seamless transition
The Stack
Ontology-driven. Source-aware. AI-native.
Not a linear stack. A pipeline that feeds intelligence. Connect, classify, explore — then reason, act, and close the loop.
REASON
TALK
IntelWeaver AI
AI
AGENTS
Governed AI
autonomous
BI
Business Intelligence
insight
▲ consumes knowledge graph
UNDERSTAND
EXPAND
System Connectors
connect
→
MODEL
Classification Engine
classify
→
EXPLORE
Knowledge Graph
discover
▲ builds understanding
ADAPT
FLOWS
Integration flows between systems. The action loop back into reality.
↔ bidirectional
▲ derives from
LEAPED ONTOLOGY v3.1.0
The single source of truth. Everything above derives from this.
foundation
Formally proven
Classification that a reasoner can verify
The ontology isn't just a taxonomy. It's a set of formal constraints that make misclassification provably wrong — not just unlikely.
Disjointness — 40 axioms enforce that every primitive is mutually exclusive. A Location cannot be an Entity. Formally.
Cardinality — Every Event has exactly one trigger. Every Asset has at most one blueprint. These aren't conventions — they're constraints.
Functional properties — One golden type per record. One timestamp per event. No ambiguity in the graph.
Bidirectional relations — Every relationship in the graph is navigable in both directions. Query from any primitive.
Validation layer — Shape constraints on every primitive. Invalid data is caught before it enters the graph.
Audit coverage — The formal structure makes regulatory queries computable rather than interpretive.
6
primitives
17
golden types
9
golden roles
21
event subtypes
40
disjoint axioms
3
system packs
100%
classification accuracy
What if every record in every system maps to just six words?