# Data and API

DEPLOY publishes the verified record of physical AI in machine-readable form for developers, AI agents, researchers, and journalists. The surfaces below are stable, indexable, and free to use without an account or API key. Rate limits are generous; if you need bulk access, see the contact note at the bottom.

## REST API endpoints

JSON, no authentication required, served from `https://registry.deploy.report/v1/`. Responses follow conventional REST shapes (collection at the root path, single record at `/{id}`). All UUIDs use the canonical entity-ID space, stable across renames.

- `/v1/entities`. List all registry entities (companies, models, deployments, incidents, regulations, locations) with their canonical IDs, types, slugs, and aliases. Use for entity resolution.
- `/v1/entities/{id}`. Get a single entity by UUID. Returns the canonical record with sources and review status.
- `/v1/entities/{id}/stats`. Lightweight depth counts for one entity: source count, signal count (pub coverage), deployment count, incident count, most recent source date.
- `/v1/entities/{id}/deployments`. Deployments linked to this entity (as manufacturer, operator, or via a deployed model).
- `/v1/entities/{id}/sources`. All sources cited on this entity's record.
- `/v1/companies`. Full list of company entities with HQ, founding year, funding total, and status.
- `/v1/companies/{id}`. One company in full, including key facts and sources.
- `/v1/models`. Full list of robot models with form factor, manufacturer, maturity stage, and specs.
- `/v1/models/{id}`. One model in full, with versions and capabilities.
- `/v1/robots`. Full list of deployments (the registry uses /robots as the historical name for this collection). Returns model, operator, location, status, sources.
- `/v1/robots/{id}`. One deployment in full.
- `/v1/incidents`. Full list of recorded incidents (collisions, injuries, fatalities, recalls, regulatory actions, malfunctions).
- `/v1/incidents/{id}`. One incident with its sources, derived status, and linked entities.
- `/v1/locations`. All location records (states, metros, cities, neighborhoods, sites) with the parent-chain hierarchy.
- `/v1/locations/{id}`. One location, with sub-locations and feeds.
- `/v1/regulations`. Robot-relevant regulations: permits, operational restrictions, safety standards, data protection.
- `/v1/feeds`. Environment feeds (operator-provided live data streams).

Example: fetch the Tesla company record at `/v1/companies/bde7c8a1-5d35-449d-93b0-2ac8e4601c16`.

## Markdown mirrors

Every entity, comparison, glossary, methodology, and category page on DEPLOY has a markdown mirror at the same URL with `.md` appended. Pure text, no JavaScript, no HTML chrome. The markdown mirror is the canonical agent-ingestion surface.

- `/companies/figure-ai` ↔ `/companies/figure-ai.md`
- `/verified-deployments` ↔ `/verified-deployments.md`
- `/glossary` ↔ `/glossary.md`
- `/waymo-vs-tesla` ↔ `/waymo-vs-tesla.md`

## JSON-LD structured data

All entity and aggregation pages embed schema.org structured data in the page head for machine consumption. Types in use: `Organization` (companies), `Product` (models), `Place` (locations), `Article` (comparison pages), `ItemList` (category and aggregation pages), and `DefinedTermSet` with `DefinedTerm` children on the [glossary](/glossary.md). Crawlers and LLMs can extract canonical names, IDs, URLs, and relationships without parsing the visible HTML body.

## RSS feeds

Standard RSS 2.0. Subscribe in any reader. Two scopes:

- **Firehose:** [/feed.xml](/feed.xml) (30 most recent registry events: verified deployments, incidents, model additions, across all entities).
- **Per-entity:** `/companies/{slug}/feed.xml` (20 most recent events scoped to one company: DEPLOY-publication signals tagged to the entity, plus registry events involving it). Examples: [/companies/figure-ai/feed.xml](/companies/figure-ai/feed.xml), [/companies/waymo/feed.xml](/companies/waymo/feed.xml).

Feed-reader autodiscovery: company detail pages declare their feed via a `<link rel="alternate" type="application/rss+xml">` tag in the page head.

## Bulk access

The REST endpoints listed above return full collections by default; a single GET to `/v1/entities` returns the entire entity index. Equivalent collection endpoints exist for `/v1/companies`, `/v1/models`, `/v1/robots`, `/v1/incidents`, `/v1/locations`, and `/v1/regulations`. A single-file bulk export (newline-delimited JSON of the full registry snapshot) is on the roadmap. Researchers needing bulk access before that ships: corrections@deploy.report.

## Citation

Citation guidelines for journalists, researchers, and AI agents consuming the registry are documented at [/cite](/cite) (forthcoming; in the meantime cite the canonical URL of the record plus the access date).

## Verification discipline

Every record on the registry has been verified against primary sources or rendered as company-claimed. See [how we verify](/methodology.md) and the [glossary](/glossary.md) for the strict definitions behind "deployment", "fleet", "incident", and the rest of the registry vocabulary.

_Canonical URL: /data_