Data and downloads
As of Nov 18, 2026, Food Robot Index publishes 218 location records, 26 announcement records, and 248 source records as open data under CC BY 4.0.
Downloads
| File | Format | What it holds | Entries |
|---|---|---|---|
| installs.csv | CSV | One row per location: status, dates, place, maker, model, sources. | 218 |
| announcements.csv | CSV | Count and plan claims that are not named locations. | 26 |
| makers.csv | CSV | Robot makers and their listed products. | 52 |
| chains.csv | CSV | Chains and operators. | 149 |
| sources.csv | CSV | Every cited public source with its date, type, archive link, and a short excerpt. | 248 |
| installs.json | JSON | Install records with full status history. | 218 |
| announcements.json | JSON | Announcement records. | 26 |
| makers.json | JSON | Maker records. | 52 |
| chains.json | JSON | Chain records. | 149 |
| sources.json | JSON | Source records. | 248 |
| changelog.json | JSON | Weekly change log. | 1 |
| foodrobotindex-full.json | JSON | One file with every table above plus a metadata block and the citation line. | 693 |
Each chain and maker page also has its own CSV of that company’s locations, at /data/chain/ID.csv and /data/maker/ID.csv. They use the same fields and values as installs.csv, trimmed to the main columns, with a license column.
The CSV files use the same field names as the JSON files. Text is UTF-8, dates are YYYY-MM-DD, empty cells mean no value, and lists are joined with a pipe (|). In installs.csv, status_history is a pipe-joined list of status@date. The full history, with notes and sources per event, is in the JSON.
Current location counts by status: 145 in use, 18 in a trial, 3 announced, 29 not yet confirmed, 23 removed.
License
The data is licensed under Creative Commons Attribution 4.0 (CC BY 4.0). You may copy, share, and adapt it, including for commercial use, if you give credit. Credit line: Food Robot Index, foodrobotindex.com.
The site code and design are not part of this data license. Short source excerpts in sources.csv and sources.json are quotes kept so each fact can be checked, and they stay with their publishers. The full text is in LICENSE-DATA.md.
How to cite
Use this line and keep the data version date:
Food Robot Index, foodrobotindex.com. Restaurant and food-service robot list, data version 2026-11-18. https://foodrobotindex.com/data/. CC BY 4.0.
The data version is the date of the newest edit in the data. It changes with each weekly update. Cite the label and date shown for a location, not a later one. Food Robot Index is a public record, not investment advice.
Data dictionary
The data uses these status values: announced, piloting (shown on the site as “in a trial”), live (“in use”), removed, and unconfirmed (“not yet confirmed”). The field names stay the same in every file. The methodology page says what evidence each label needs. Map points at a city or state center have geo_precision set to city or region and are shown as “roughly placed.”
Install records
| Field | Type | Meaning |
|---|---|---|
id | string | Stable slug. Pattern: {brand-slug}-{location-slug}-{model-slug}. Example records use the prefix EXAMPLE-. |
brand | string | Chain or independent operator as the public would name it. |
location | string | Display name of the specific site (brand + store/site name). |
robot_maker | string | Manufacturer name. |
model | string | Product name (for example, a fry-station model). |
task | string | What the robot does, in plain language (fry station, bowl makeline, drink making, in-restaurant delivery, and so on). |
status | enum | One of the five status values above. |
first_seen | date | Date of the earliest public signal for this location-level record (YYYY-MM-DD). |
first_seen_basis | enum (optional) | reported (default) | recorded. Use recorded when every source for the row is undated (a store locator or case study with no publish date). Then first_seen is the date the ledger saw the page, not the date the site opened. The site shows these rows from that date and leaves them out of the timeline bar chart. |
status_since_basis | enum (optional) | Same values. Used when status_since is a recorded date. |
last_confirmed | date | Date of the latest source that supports the **current** status. Alias: last verified. |
source_links | string[] | IDs of records in sources.json. At least one is required. |
notes | string | Public, factual notes. Empty string if none. |
confidence | enum | high | medium | low. See rules. |
brand_id | string | FK to chains.json. |
maker_id | string | FK to makers.json. |
operator | string | null | Franchisee or contract operator if different from brand. |
robot_category | enum | Controlled task family for filters and the scoreboard. Values: fry_station, grill, bowl_line, pizza_line, drink_maker, prep (for example avocado prep), in_restaurant_delivery, bussing, dishwashing, other. |
unit_count | integer | null | Units of this model at this site, if a source states a number. |
address | string | null | Street address only if a public source or public store locator publishes it. Do not guess. |
city | string | null | |
state | string | null | Region/state/province. |
postal_code | string | null | |
country | string | ISO 3166-1 alpha-2. |
lat | number | null | WGS84. Null if not geocoded. |
lng | number | null | WGS84. Null if not geocoded. |
geo_precision | enum | address | venue | city | region | country | approximate | unknown. Describes the coordinate, not the site. A city, region, or country coordinate is a center point and must never be presented as a precise pin. The site shows these as approximate. |
geo_basis | enum (optional) | source (default) | geocoded_city_center | geocoded_region_center. The geocoded values mean we looked up the middle of the named city or state on OpenStreetMap Nominatim because no source gave coordinates. Cache: logs/geocode-cache.json. Script: process/geocode-cities.py. |
location_type | enum | restaurant | food_hall | campus | hospital | stadium | airport | ghost_kitchen | c_store | hotel | other. Used to enforce scope edges. |
status_since | date | Date the current status began (may be earlier than last_confirmed). |
first_announced_date | date | null | Date this location was first publicly announced, if known. |
needs_reverification | boolean | True when the staleness window has lapsed or a conflicting signal appeared. Does not by itself change status. |
status_history | array | Ordered events; see below. The change log is derived from this. |
last_updated | date | Date this record was last edited in Food Robot Index. |
is_example | boolean | Must be true on template rows; omit or false on real rows. |
Status history events (inside each install)
| Field | Type | Meaning |
|---|---|---|
status | enum | Status after the event. |
date | date | Date of the underlying evidence, not the date we typed it in. |
recorded_on | date | Date we captured the event. |
source_links | string[] | Sources that justify this transition. |
note | string | Short public note. |
date_basis | enum (optional) | reported (default) | recorded. recorded means date is the day we saw the evidence (for example, a listing gone from a locations page), not a date a source states. |
Announcement records
| Field | Type | Meaning |
|---|---|---|
id | string | Stable slug, prefix EXAMPLE- on templates. |
claimant_type | enum | maker | brand | other. |
claimant | string | Who made the claim. |
claimant_id | string | null | FK to makers.json or chains.json. |
brand | string | null | Brand named in the claim, if any. |
brand_id | string | null | |
maker_id | string | null | |
model | string | null | |
task | string | null | |
robot_category | enum | null | |
scope | enum | unit_count | location_count | chain_wide | region | partnership | other. |
claimed_unit_count | integer | null | |
claimed_location_count | integer | null | |
geography | string | null | Free-text geography as stated (“United States,” “Midwest,” and so on). |
announced_date | date | |
target_date | date | null | Stated deadline or “by year” date, if any. |
status | enum | active | revised | withdrawn | fulfilled | expired. |
claim_text | string | Short paraphrase of the public claim, not a slogan. |
source_links | string[] | Required. |
supersedes | string | null | ID of a prior announcement this revises. |
confidence | enum | high | medium | low. |
notes | string | |
last_confirmed | date | |
last_updated | date | |
is_example | boolean | |
claim_kind | enum | null | Optional. reported = the company's own count of units or sites in operation as of a stated date (for example Sweetgreen's Infinite Kitchen count). Absent or plan = an announced plan or target. The site says "reported" for the first and "announced" for the second. |
Source records
| Field | Type | Meaning |
|---|---|---|
id | string | Stable slug. |
url | string | Canonical public URL. |
publisher | string | |
published_date | date | null | Null only if the page is undated. |
accessed_date | date | Date we last opened the public page. |
archive_url | string | null | Wayback (or equivalent public archive) URL. Required before a status upgrade when the archive succeeds. |
source_type | enum | first_party_maker | first_party_operator | press_wire | trade_press | local_news | national_news | ir_or_sec | earnings_call | official_social | user_social | review | job_posting | photo_video | other. |
quote_excerpt | string | Short excerpt that supports the fact cited. Keep it short. |
paywalled | boolean | If true, prefer a free corroborating source as well. Do not create an account to bypass a paywall. |
notes | string | |
is_example | boolean |
Maker and chain records
Makers: id, name, website, newsroom_url, products (list), notes, in_v1_scope. Chains: id, name, website, newsroom_url, notes. They are directories of public organizations, not install claims.
Credits
The near me page: Typed ZIP codes are placed using a table of 3-digit ZIP prefix centers worked out from the GeoNames postal code data for the United States, licensed CC BY 4.0. A prefix center can be 10 to 40 miles from the exact ZIP.
For search tools and AI assistants
- llms.txt: an index of key pages, data files, and rules.
- llms-full.txt: one plain-text line per location, plus every announcement and scoreboard sentence.
- sitemap.xml and the Atom feed list every page and weekly update.
Updates are weekly. Crawlers, including AI crawlers, are welcome. See robots.txt.
Last updated: Nov 18, 2026.