Quick start
The fastest way to start working with the Workforce Labor Index is to clone the public repository. The repo contains the full methodology document, the JSON snapshot of current category data, the eval harness configuration, JSON-LD schemas used across the public site, and the LaTeX source for the citation entry.
Clone the repository:
# Clone the Workforce Labor Index git clone https://github.com/griffainai/workforce-labor-index.git cd workforce-labor-index # Inspect the machine-readable snapshot cat data/categories.json | head -40 # Read the methodology source open docs/methodology-v1.0.md
The repository ships with a stable directory layout. The data/ directory holds the current snapshot and dated historical exports. The docs/ directory holds the methodology document, the IOSCO PD415 statement of compliance, and the eval harness specification. The schemas/ directory holds the JSON-LD definitions that appear in <script type="application/ld+json"> tags on every public WLI page.
No authentication, no API key, no rate limit. The repository is the substrate — everything WorkForce publishes about the index ultimately resolves to a file in this repository or a section in the methodology document.
What’s published
The WLI is a benchmark, not a black box. Everything required to reproduce a category figure independently is published. The surfaces below are the canonical sources; any rate, AQO score, or methodology claim that appears in WorkForce’s marketing, reports, or marketplace ultimately resolves to one of them.
Public GitHub repository
github.com/griffainai/workforce-labor-index is the source of truth. It contains the JSON snapshot, the methodology document, schemas, and historical exports. Fork it, file issues against the methodology, or open a pull request with a category proposal. The repository is the contract.
WLI Methodology v1.0
The full calculation pipeline is published at /methodology. It documents how categories are defined, how observations are weighted, how the AQO score is composed, how the eval harness is sealed, and how each version is frozen. A versioned PDF and Markdown source sit in the repository under docs/methodology-v1.0.md.
IOSCO PD415 Statement of Compliance
WorkForce publishes a Statement of Compliance against the IOSCO Principles for Financial Benchmarks (PD415). The statement is at /wli/iosco-compliance with the source under docs/iosco-pd415.md. This is the document insurers, regulators, and procurement teams ask for first.
Eval harness
The sealed task bank methodology is documented at /eval. Test banks are versioned v{X.Y}; once sealed, a bank version does not change. Vendor submissions are scored against a bank version, and the version is part of every AQO citation.
JSON-LD schemas
Every public WLI page emits a JSON-LD payload — a Dataset on /wli/[category], a TechArticle on the methodology, and an Organization on the root. The schemas live in schemas/ and are designed to be consumed by AI search citation pipelines and procurement scrapers alike.
Machine-readable snapshot
The current state of the index is exported as a single JSON file at data/categories.json. Each category record carries identifiers, units, the methodology version it was computed under, the eval bank version, and provenance metadata. Snapshots are timestamped and historical exports are retained in data/snapshots/.
What you can build
The license is permissive on purpose. The four patterns below are the most common ways early integrators are working with the published data and methodology. None of them require contacting WorkForce or signing an agreement — the methodology is the contract.
Procurement integration
Pull the JSON snapshot into a procurement workflow to anchor vendor pricing against the published median for a category. A common pattern is a nightly fetch of data/categories.json from main, joined against an internal vendor table by category identifier. The result is a procurement dashboard that surfaces vendors whose rates deviate materially from the WLI median, with the methodology version cited inline so the comparison is auditable.
FinOps pipeline
FinOps teams use the index to reconcile AI labor spend the same way they reconcile cloud spend against published SKU pricing. Categories are stable identifiers; the methodology version is a stable reference; the JSON-LD Dataset schema is designed to be ingested directly into a data warehouse without further transformation. Pin a methodology version in the pipeline and the entire reconciliation becomes reproducible quarter over quarter.
Academic reproduction
The methodology, the sealed eval banks, and the historical snapshots are sufficient to reproduce a category figure from first principles. Cite the index as a dataset using the BibTeX entry under CITATION.bib, pin the methodology version and snapshot date in the paper, and link to the GitHub commit hash. Reproductions are welcome — file them as issues in the repository.
Marketplace listing
Vendors who want to list on the WorkForce marketplace submit through the eval pipeline. The published harness specification documents the input format, the sealed bank a submission will be scored against, and the AQO composition. A vendor can run the harness locally to estimate its score before submitting. The marketplace listing carries the eval bank version, the methodology version, and a permanent link to the submission record.
API roadmap
This is the honest picture of the developer surface today and where it goes next. The repository is the primary integration surface right now; programmatic endpoints arrive with transaction-one of the marketplace. We do not publish mock responses or fake endpoints.
data/categories.jsonFull current state of the index, fetched from GitHub raw or the cloned repo.docs/.docs/eval-harness.md.Until the REST surface ships, the canonical integration pattern is a scheduled fetch of the JSON snapshot from main, with the methodology version and commit hash pinned in the consuming system. This is the same pattern researchers use to cite a frozen dataset, and it carries over cleanly when REST endpoints land — the URLs change, the contract does not.
Methodology versioning
Every figure the WLI publishes is bound to two version identifiers: the methodology version (the calculation pipeline) and the eval bank version (the sealed task bank used to score AQO inputs). Both are append-only. A methodology version, once published, is never edited in place; revisions become a new version.
Cite a specific version of the WLI in a paper or report using the entry below. Replace methodology_version and snapshot_date with the values you pinned.
% BibTeX — citing a specific version of the WLI
@misc{workforce_labor_index_v1,
title = {Workforce Labor Index},
author = {{WorkForce}},
year = {2026},
howpublished = {\url{https://workforcebygriffain.com/methodology}},
note = {Methodology v1.0, snapshot YYYY-MM-DD,
eval bank v{X.Y}, commit <sha>},
license = {CC-BY-4.0}
}In running text, the recommended form is: Workforce Labor Index, methodology v1.0, snapshot YYYY-MM-DD (WorkForce, 2026). The methodology document and the snapshot file are both retrievable from the repository at the commit hash; this is sufficient for independent reproduction.
The versioning contract is deliberately conservative. A methodology version bump is a publishing event — documented in the changelog, called out in the newsletter, and recorded in the ledger. A snapshot revision is a routine event and lands as a commit. Consumers who pin both values get reproducible figures indefinitely.
License
The Workforce Labor Index — the methodology, the snapshot, the schemas, the eval harness specification — is licensed under the Creative Commons Attribution 4.0 International License (CC-BY-4.0). This is a deliberate choice. A benchmark only becomes a reference when it can be cited without permission.
Under CC-BY-4.0 you may: share — copy and redistribute the material in any medium or format; adapt — remix, transform, and build upon the material for any purpose, including commercially. You may build paid tools on top of the WLI. You may fork the methodology and publish derivative work. You may embed WLI figures in proprietary procurement systems.
The single requirement is attribution: give appropriate credit, link to the license, and indicate if changes were made. In practice this means citing the methodology version and the snapshot date, and linking to github.com/griffainai/workforce-labor-index. The published BibTeX entry covers this in formal contexts; a plain-text citation is sufficient in informal ones.
- Cite — use any published rate, AQO score, or methodology document in commercial or academic work, with attribution.
- Fork — fork the repo, run a derivative calculation, publish your variant. Attribution required, share-alike not required.
- Build — wrap the JSON snapshot in your tool, your dashboard, your procurement workflow. No license fee. No usage cap. No permission email.
- Attribute — include the BibTeX entry above, or link to the methodology page with the version tag. That is the full attribution requirement.
There is no separate commercial license, no enterprise tier of the methodology, no closed-source variant. The methodology is the public substrate the marketplace operates on; commercial value accrues at the transaction layer, not at the reference layer. This is the part of the architecture that is not for sale.