Practice overview
Information designed to retain its meaning
We design information environments that make every resource intelligible to people and systems. Users can see what a resource represents and where it came from. They can also follow its relationship to other evidence and judge whether it remains suitable for the task at hand.
Our work begins with a conceptual model and the metadata needed to express it. Vocabularies establish shared meaning, while dependable pipelines carry that meaning into data products and knowledge graphs. Governance turns the architecture into a working practice. Local systems can retain necessary distinctions because shared identifiers and definitions provide connection without forced uniformity. Provenance and quality controls support reuse across the organisation.
Method
How we structure the work
We select methods around the information involved and the decisions it must support. Four connected layers keep the work grounded in operating constraints.
Discover
We start with the priority questions and trace the resources used to answer them. This reveals how systems support the workflow and where terminology diverges. We also establish who uses the information and who has authority over it. Profiling representative evidence then exposes failures of structure or meaning, along with weaknesses in quality and access. The governance gaps become visible through the same analysis.
Model
We define the concepts that the target uses depend on and make their relationships explicit. Stable identifiers and metadata allow those concepts to travel between systems. Constraints state what valid information should look like, while provenance preserves its origin. Existing standards are reused where they fit and extended where necessary.
Engineer
We design how information enters the environment and how it is transformed. Reconciliation preserves identity across sources before validated information reaches storage. Search and access patterns are built around real user needs. Each architecture choice reflects the character and scale of the sources. Security requirements are balanced with latency, maintainability and integration needs.
Govern
We assign stewardship and make the boundaries of authority clear. Versioning preserves the history of change, while quality measures show whether the information remains fit for use. Change control is integrated with the wider lifecycle. Adoption is judged by practical outcomes: can people find information, connect it across systems and trace it back to source? Evidence of reuse shows whether the environment is becoming part of ordinary work.
Capabilities
Capabilities within the discipline
We combine these capabilities around how the organisation represents and uses information. The same architecture connects its data with the wider knowledge system.
Metadata & Interoperable Data Systems
We make information exchangeable while preserving the differences that affect interpretation. Its structure and units remain explicit. The level of detail stays tied to the original context, and the rules governing updates remain visible.
Including
- Statistical schema induction
- Schema composition, alignment, and mapping
Taxonomies, Ontologies & Controlled Vocabularies
We establish governed concepts and the labels through which people encounter them. Explicit relationships provide context, while constraints prevent invalid interpretations. The resulting language supports consistent classification and better discovery. It also gives integration and analysis a shared semantic foundation.
Including
- Ontology alignment, evaluation, and design
- Structured, unstructured, and multimedia resource ingestion
Semantic Layers & Application Profiles
We create a stable layer of shared meaning above source systems and specify how broader standards apply in a particular operational context.
Including
- Semantic-layer maturity assessment and ontology grounding
- Decision-trace auditing and provenance
- Ontologically grounded agentic AI
Knowledge Graph Systems
We connect the entities that matter to the evidence held about them. Events show how that picture changes over time. Rules make valid interpretation explicit and provenance retains the route back to source. This approach is valuable when relationships are central to the questions being asked.
Including
- Business-domain and enterprise knowledge graphs
- Graph-based logic induction and enrichment
- Multi-graph orchestration
- Hybrid neurosymbolic and LLM systems
Enterprise Knowledge Management
We make expertise and institutional memory available in the course of work. Evidence remains connected to the procedural knowledge that gives it context. Clear ownership helps people trust what they find and keeps it maintained as practice changes.
Including
- Organisational silo assessment and reconciliation
- Tacit, tribal, and procedural knowledge capture
Data Engineering
We build dependable data products on pipelines whose transformations can be inspected. Explicit lineage preserves their relationship to source. Quality controls are chosen for the intended use and embedded in an operating process that teams can maintain.
Including
- Aggregation, normalisation, cleaning, and deduplication
- Entity resolution and labelling
- Applied statistics and data modelling
- Advanced transformation and enrichment
- Data quality, integrity, and completeness auditing