The physical world before it becomes a product, record, label, or system entry.
The Data Preparation Layer
Do not cook on the delivery truck.
This page uses a food supply chain analogy to explain why trusted data platforms need a preparation layer between source systems and data consumers.
Different data sources are like ingredients from different origins. They come with different owners, formats, assumptions, freshness, and quality.
A trusted data platform works like a professional kitchen. It prepares ingredients once, with quality checks, labels, identity, storage rules, provenance, and governance, so many dishes can be made without repeating the same preparation work.
The semantic layer and ontology act like recipes and cooking logic. They explain what ingredients mean, how they can be combined, and what rules apply.
AI, applications, analytics, and simulations should not need to search every source system for raw ingredients. They consume prepared, governed data products to create trusted decisions and business outcomes.
RealityCaptureIngredientsKitchenPrepared Data ProductsRecipesMealsDecision Value
Core principle
Curate once, reuse many.
Separate data production from data consumption. Source systems should not become places every application, analyst, or AI agent has to search independently.
The platform layer turns raw and semi-prepared data into governed data products, semantic definitions, knowledge assets, and decision-ready outputs.
Do not cook on the delivery truck.
Source systems and data pipelines are not the right place for every application, dashboard, or AI agent to prepare its own data. A trusted data platform works like a modern kitchen: it receives raw and semi-prepared ingredients, checks quality, records provenance, standardises preparation, applies governance controls, and serves reusable data products to many consumers.
Reality and source systemsReality CapturePrepared Data ProductsSemantic LayerAI, applications, and decisions
Reality Capture as Food Preparation
Different capture methods create different prepared representations of the same reality.
Before food reaches a kitchen, it has already gone through a supply chain: harvesting, catching, collecting, cleaning, cutting, packaging, transporting, storing, and labelling.
Data has a similar lifecycle. Reality is captured through sensors, GIS, BIM, LiDAR, imagery, documents, surveys, operational systems, and human records. Each capture method creates a different digital representation, with its own assumptions, precision, freshness, coverage, and intended use.
For spatial and digital twin platforms, this is especially important. The same real-world object may appear as a map feature, BIM element, point cloud, sensor record, document reference, scene object, or knowledge graph node. Governance connects these representations back to the same reality through identity, metadata, lineage, quality, semantic mapping, and observability.
GIS featureBIM elementLiDAR point cloudImagerySensor recordDocument referenceScene objectKnowledge graph node
Lifecycle analogy
From source ingredients to meals and nutrition.
The analogy is strongest when it follows the full lifecycle. Source systems provide raw ingredients; capture and preprocessing create prepared representations; the middle platform governs them; semantic logic explains how to combine them; AI and applications consume trusted outputs.
Source IngredientsSource Data
Different systems produce data with different owners, formats, assumptions, freshness, and quality.
Capture and HandlingSource ProcessingData is captured, transformed, packaged, and moved before it reaches the platform. Those upstream assumptions need to be understood.
Market and StorageData InventoryThe platform needs to know what data exists, where it came from, who owns it, and whether it is fit for use.
Professional KitchenData Middle PlatformPrepare trusted data once so many consumers can reuse it safely.
PreparationData Quality, Metadata, IdentityClean, classify, identify, validate, and trace data before it is reused.
Recipes and Cooking LogicSemantic Layer and OntologyDefine shared meaning, relationships, rules, and constraints for consistent reuse.
Meals and NutritionDecisions and OutcomesAI, applications, analytics, and simulations consume trusted data to create decision value.
Teaching scenario
What a global food system can teach us about trusted data.
Data Kitchen is not a food framework. Food is a teaching model for explaining why repeatable decisions need standardisation, governance, supply-chain visibility, quality controls, metadata, and shared meaning.
A McDonald's-style operating model is useful because the product looks simple to the customer, but behind it sits a governed system of suppliers, recipes, labels, preparation standards, quality checks, version control, and impact analysis.
Different systems represent the same real-world ingredients in different ways.
The same item or batch needs persistent identity across systems and handoffs.
Data becomes reusable only after preparation, quality checks, metadata, and control.
Business meaning is defined independently from individual source ingredients.
Relationships connect products, ingredients, suppliers, standards, metadata, and decisions.
Trusted data supports impact analysis, operational response, and business decisions.
Great decisions are not made from raw ingredients.
They are made from well-prepared, trusted recipes. Ingredients are data. Recipes are semantic definitions, governance rules, and business logic. Meals are decision assets.
One ingredient set, many recipes
The same prepared ingredients can serve very different end uses.
A simple chain burger may look like one product to the customer, but behind it sits a governed system of suppliers, ingredients, preparation standards, labels, quality checks, recipe versions, and regional variations.
Data products work the same way. A dashboard metric, AI answer, digital twin view, risk score, or decision recommendation may use overlapping prepared data ingredients, but each end use needs its own recipe: definitions, quality thresholds, relationships, rules, context, and governance controls.
Implementation can vary. Meaning must be governed.
The semantic layer and ontology provide that governed, reusable meaning. They explain what the data means, how it can be combined, which rules apply, and how changes affect downstream decisions.
Universal teaching model
Kitchen Methodology can explain trusted data across industries.
The point is not the food example itself. The point is to give business, data, architecture, operations, and AI teams a shared language for why preparation, governance, semantics, and feedback loops matter.
McDonald's-style food system
Data governance, semantic layer, metadata, quality, lineage, standardisation
Airport operations
Reality mapping, identity, operational events, service dependencies
Digital twin city
Source-to-scene, spatial relationships, scene governance, knowledge graph
Banking operations
Master data, lineage, controls, regulatory evidence, decision accountability
Warehouse fulfilment
Data products, event streams, automation, quality, operational feedback
Consulting question
Are teams consuming prepared data products, or are they still rebuilding ingredients from source systems?
The answer reveals whether the organisation has an AI-ready data platform or a growing collection of fragile point solutions. Prepared data products are governed, reusable data assets that have already been cleaned, described, identified, quality-checked, linked, and made ready for repeated consumption.
Where Data Is Different from Food
The analogy helps, but data needs stronger governance than food.
Food is consumed when it is used. Data can be reused, copied, linked, enriched, and improved. With the right feedback loops, data quality can increase over time as issues are detected, definitions are clarified, lineage is strengthened, and semantic mappings are improved.
Food can often be cut into smaller pieces. Data granularity is limited by how reality was captured and how source systems recorded it. A building-level record cannot automatically become a trusted room-level or component-level record without new evidence, inference, or modelling.
Data supply chains can also become longer than food supply chains: source systems, transformations, matching rules, semantic mappings, knowledge graphs, vector indexes, AI agents, decisions, and feedback loops. This makes lineage, provenance, versioning, ownership, observability, semantic controls, and decision accountability essential.
FoodData
Consumed when usedReused, copied, linked, and enriched
Quality often declines over timeQuality can improve with feedback
Can be physically dividedGranularity depends on capture and source systems
Supply chain ends at consumptionUsage creates feedback and new data
Freshness is centralFitness for purpose is central
Recipes guide cookingSemantics and ontology guide reuse
Food safety protects consumersGovernance and observability protect decisions
Connected to the concept network
This page in the Trusted Data Framework.
Explains the operating model for curating once and reusing many times across AI, applications, and decisions.
Raw DataCurated DataSemantic LayerKnowledgeIntelligence
Related concepts
Consulting assets
Data Product Contract TemplateAI-Ready Data Foundation Assessment
Delivery connection
DesignGovernOperate
Cross-cutting
GovernanceQualityReuse