Structured knowledge, stable identifiers, open data
NEI is designed to be machine-readable from the ground up. Every indicator has a stable identifier, a structured data representation, and a predictable URL. Build on it with confidence.
The data model
The framework has four primary data types, each with a stable identifier and a machine-readable representation:
Concept files
The permanent definition of an indicator — its ID, canonical title, and domain. Never changes once assigned.
Version files
The versioned specification — assessment criteria, evidence requirements, supporting and dissenting citations.
Taxonomy nodes
Domain definitions with stable IDs, descriptions, and edges to parent/child domains.
Release manifests
A snapshot of the framework at a point in time — which indicators, at which versions, were released together.
Stable identifiers
NDI identifiers are generated by a deterministic algorithm: normalize the indicator's
canonical title → SHA-256 hash → base32 encode → lowercase → take first 6 characters → prefix NDI-.
The same title always generates the same ID. This means IDs can be verified independently
and are not dependent on a central registry.
Once assigned, an NDI identifier is permanent. It is never reassigned, never reused,
and never retired in a way that makes the original reference invalid. This makes NDI
identifiers safe to use as foreign keys in databases, as citation targets in research,
and as stable references in software integrations. A reference to NDI-hgbbzn-v1
means exactly the same thing today as it will in ten years.
Machine-readable data
Structured data is available at stable URLs in JSON format:
| Endpoint | Description |
|---|---|
| /data/nei-latest.json | Full framework data for the current release — all indicators, taxonomy, and metadata |
| /data/nei-mini.json | Compact, LLM-optimised representation — indicator IDs, titles, and one-line descriptions |
| /data/indicators/{id}.json | Single indicator data — concept and latest version specification |
| /data/taxonomy/{id}.json | Single taxonomy node — domain metadata and edges |
AI and LLM integration
NEI indicators are designed as retrievable knowledge chunks. Each indicator has a
stable ID, a short title, a one-paragraph description, and structured assessment
criteria — a format that works well for retrieval-augmented generation and for
structured extraction tasks. The /data/nei-mini.json file is optimised for
inclusion in system prompts.
A structured output schema is available for mapping text — job descriptions, policy documents, employee reviews — to NEI indicators. AI systems can use this schema to tag content with relevant indicator IDs, enabling downstream analysis against the framework. The LLM usage guide has example prompts and integration patterns.
Standards as code
The framework is developed in a public Git repository. Source files are YAML and CSV. Every change is a commit with a documented rationale. The build process generates JSON data files and human-readable documentation from the same source. If you want to understand exactly how an indicator was defined or changed, the full history is in the repository.
Build on NEI
What you can build with stable identifiers, structured data, and an open license:
Analysis tools
Tag organisational content against indicators. Score organisations against the framework. Generate reports from structured evidence.
HR software integrations
Embed NEI indicators in job description tools, policy management systems, or performance review platforms.
AI assistants
Build assistants that can answer questions about workplace design using NEI indicators as a structured knowledge base.