API Automations runs a research program alongside its products. Public, licensed and consented data is normalized into source-attributed, aggregate signals under one governance standard, and every result is published with the method, the data and the disclosures that produced it.
Federal health agencies publish a great deal of data that is rarely analyzed at the moment it would matter: adverse-event reports, device clearances and recalls, registered trials and their results, and the public comments that shape the guidance those agencies issue. This track takes a question a clinician actually has, builds a reproducible, dated extract of the public data that bears on it, and puts the answer where it can change something: a regulatory docket first, then the peer-reviewed literature.
Clinical outputs are led by each project’s clinical lead; data outputs by API Automations. Code, data and disclosures are published alongside every result.
Explore the health track →Every project on this page sits in one of four domains. The domain fixes the question; the pipeline and the gates further down fix how it is answered.
How consumer devices, CTV platforms and advertising SDKs transmit data across networks. Telemetry mapping identifies opaque behaviors and supports privacy-centric design standards.
The transparency and accessibility of institutional APIs (civic, governmental and regulatory datasets), studied to inform equitable, accountable data governance. Active projects: PEPI Index, Estate Index, Cannabis Research Index & Civic Forecast.
A separate track: reproducible analyses of public federal health data, from adverse-event surveillance to regulatory dockets. Clinical outputs are led by clinicians and data outputs by API Automations, published with the code, the data and the disclosures.
Read about the trackTechnical findings turned into public education through Data Privacy Today, including the “What You Need to Know in 90 Seconds” series.
Each output comes from the same pipeline. The indexes are computed from public filings and official statistics, the model answers against sourced records, and the papers are the analyses that went to a docket or a journal. Every output links to its live site; a gated one opens on its own sign-in and says so on the card.
Every number here names where it comes from. The Cannabis Research Index figures are copied, with their date, from the index’s own source register; the health figures are computed at build time from the dated registers the project keeps, so a chart cannot drift from its file.
| Category | Count |
|---|---|
| Not applicable | 10 |
| Phase 4 | 8 |
| Phase 3 | 1 |
| Phase 1 | 1 |
| Not stated | 1 |
ClinicalTrials.gov registry snapshot · snapshot 2026-09-29
| Category | Count |
|---|---|
| 2014 | 1 |
| 2016 | 1 |
| 2017 | 1 |
| 2019 | 1 |
| 2021 | 1 |
| 2022 | 1 |
| 2025 | 1 |
| 2026 | 2 |
Project literature register, verified · snapshot 2026-09-29
| Category | Count |
|---|---|
| United States | 21 |
| Canada | 8 |
| International bodies | 2 |
| Research registries | 2 |
| Uruguay | 1 |
| Germany | 1 |
| Portugal | 1 |
| Switzerland | 1 |
| Australia | 1 |
Cannabis Research Index source register · snapshot 2026-09-09
Registration is not collection. Every family here has been assessed and admitted to the register, and 5 of them supply observations today, the largest being Canada’s national statistics agency. The rest are listed so the gaps are visible rather than implied.
The register is the index’s own, published on its coverage page. Values, period-level history and series definitions are licensed rather than published, so none of them appear here.
Figures for the access-controlled products are not shown here. They are reported to licensed users under the terms of their agreement rather than published to the open web.
Photo by Steve A Johnson on Unsplash
Governance is a claim about how something is run, so here is the run: the pipeline every source moves through, the boundaries on what a supplier’s data can become, the commitments made to every provider, and the gates a manuscript passes before it leaves. Open each for the detail.
Authorized API access or licensed feeds, with caching, retries, and rate-limit compliance, and only ever through a provider's sanctioned method and credentials.
Mapped into internal schemas with source provenance preserved as a first-class field rather than reconstructed later from memory.
Joined to the stable geographic and entity identifiers each source supports, so records aggregate without record-level exposure.
Aggregates, deltas, rates, per-capita values, severity bands, and topic tags. Transformative indicators, not the source records.
Indices, cards, and curated delivery to licensed destinations, with source attribution carried onto every surface it lands on.
Aggregate usage only: counts, rates, cohorts, and geography or topic buckets. No user-level export; partner reporting is aggregate-only.
We do not republish source content as journalism, and we do not hold ourselves out as a newsroom.
We operate a licensed marketplace, so data does move through us. What never moves is a record outside the terms it was licensed under, or to a recipient its consent does not cover.
We do not reconstruct a provider's dataset, offer a queryable record store, or build a competitor to their product.
No cross-app tracking underwrites anything we build. Consent travels with the record, and enforcement happens inline.
The full sourcing and attribution practice is set out in Open APIs & Public Data.
Every input is named and classified. A restricted input stops the paper.
A source, a journal, and an index are three different things, and only one of them accepts a manuscript.
We hold no review board of our own, so a determination is obtained from one qualified to make it, never assumed.
Cleared per source, including the separate act of publishing a figure derived from it.
The commercial structure behind the work, and any industry payments to the authors, are both declared.
The cleanest route through the ethics gate is an academic collaborator whose institution acts as review board of record. That is a real dependency rather than a courtesy, and we would rather say so. Partners get a documented method, a governed pipeline, and a corpus that is already computed, not a data dump to clean.
Empirical analysis combined with system-level inspection. Every experiment runs in a controlled, consented data environment and is documented for peer review and open publication.
We work with institutes, clinicians, journalists, and industry partners advancing data ethics, privacy innovation, and API standardization.
