The World's Source of Truth
for Food Safety
Live government data sources across the US, UK, Canada, France, and the Netherlands — queried directly at request time. AI-assisted research for additional jurisdictions where public records are accessible. For everywhere else, SafeEats™ links to the official health department portal so you can search manually. We don't call that coverage — it's a redirect, not the invention.
Why SafeEats™ Exists
Food safety data should be accessible — not buried in county portals, PDF archives, or behind paywalls. SafeEats™ aggregates inspection records from official health departments across five countries and presents them through a single, consistent interface with a universal A–F grading system.
Every jurisdiction grades differently — penalty points in LA, letter grades in NYC, star ratings in London, pass/fail in Chicago. SafeEats™ normalizes all of them to a universal 0–100 scale and A–F grade. Apples to apples, wherever we have data. For jurisdictions where no scrapable data exists, we link to the official portal — but that's a convenience, not coverage. The invention is the normalization; where there's nothing to normalize, there's no invention to claim.
The Universal A–F Grade
Raw scores from each jurisdiction (penalty points, pass/fail outcomes, letter grades) are normalized to a universal 0–100 scale. A score of 85 means different underlying criteria in Los Angeles vs. Chicago — the grade gives you a consistent at-a-glance verdict, but always review the full violation history for context.
Pass / Fail Jurisdictions
Some jurisdictions — like Chicago, Delaware, Toronto, and others — use a simple Pass / Fail system instead of numeric scores. A restaurant that passes gets a teal P badge, meaning it met health inspection standards. A restaurant that fails gets a red F badge, indicating serious violations that failed to meet minimum requirements.
Unknown — What does this mean?
A U grade means we found this establishment but have no official inspection records on file. This is more common than you might think — many restaurants, food trucks, and pop-ups have never been inspected, or their records aren't publicly available yet.
- • If past inspections exist, you'll see their full history and trend chart on the detail page — even if a current score can't be calculated.
- • If no inspections exist at all, the listing shows a clear "No official records found" notice — not a fabricated score.
- • A U grade is not necessarily bad — but it means you should call ahead or check with your local health department before dining.
How Normalization Works
Proprietary MethodEvery jurisdiction grades differently — penalty points in LA, letter grades in NYC, star ratings in London, pass/fail in Chicago. SafeEats™ normalizes all of them to a single 0–100 scale using a proprietary conversion methodology tailored to each source's native data structure. The process accounts for the unique scoring scheme of each jurisdiction and is applied uniformly after conversion to derive the letter grade (A≥90, B≥80, C≥70, D≥60, F<60). The specific conversion formulas, weight assignments, and tiered mappings are proprietary to SafeEats™ and are not publicly disclosed.
Current Grade vs Legacy Grade
Trend IntelligenceA single snapshot score can be misleading — a restaurant that scored 95 on its last visit might have averaged 72 over the past three years. SafeEats™ computes two distinct gradesfor every establishment with inspection history, and visualizes them on an interactive trend graph:
Current Grade
Most Recent Inspection
Derived from the single most recent inspection on file. This is what the restaurant scored today — the grade shown on the search card and in the hero section of the detail page.
Legacy Grade
All-Time Historical Average
The average of every inspection score on record, converted to a letter grade. This reveals the establishment's long-term safety pattern — not just a single good or bad day.
📈 Interactive Trend Graph
On each restaurant's detail page (when 2+ inspections exist), SafeEats™ renders an interactive area chart showing the score trajectory over time, with color-coded data points, grade threshold reference lines, and an automatic trend badge (Improving / Stable / Declining) that contextualizes the restaurant's safety direction over time.
Why this matters: A restaurant with a Current Grade of A but a Legacy Grade of C may have just had a single good inspection after years of problems. Conversely, a Current Grade of C with a Legacy Grade of A may indicate a recent slip from an otherwise strong track record. This dual-grade system gives diners context that a single score cannot.
How We Source Data
Three tiers — live APIs where they exist, AI-assisted research where records are public, and honest portal redirects where neither is possible.
Live Government APIsREAL-TIME
For jurisdictions that publish open-data APIs — including New York City, New York State, San Francisco, Chicago, King County (Seattle), Austin, Dallas County, Montgomery County MD, Delaware, Boston, Houston, Los Angeles County, Wake County (Raleigh, NC), Louisville KY, Las Vegas (SNHD), Toronto (Canada), Vancouver (Canada), the UK Food Standards Agency, France (Alim'confiance / DGCCRF), and the Netherlands (NVWA) — SafeEats™ queries the official database directly at request time. Data is current as of the moment you search.
AI-Assisted ResearchADDITIONAL JURISDICTIONS
For jurisdictions without a live API but with publicly accessible inspection records, AI reads official health department websites and public portals. GPT-5 Mini identifies restaurants; Gemini 3 Flash pulls inspection data from official sources. Results are clearly labeled as AI-estimated — never fabricated.
Multilingual Label ScanningANY LANGUAGE
Point the camera at any food packaging or sign in Japanese, Chinese, Korean, or any script. SafeEats™ translates ingredients, allergens, expiration dates, and dietary flags into plain English instantly.
Portal Redirect (Not Coverage)HONEST SCOPE
For jurisdictions with no scrapable data and no publicly accessible records, SafeEats™ links to the official health department portal so you can search manually. This is a convenience — not coverage. We do not claim to cover jurisdictions where we can only send you elsewhere. The normalization invention requires data to normalize; where there's nothing to normalize, there's no invention to claim.
We Don't Use Yelp. Not Even a Little.
BY DESIGNYelp is a consumer review platform. Your food safety intelligence should not come from the same pool that includes one-star reviews because the parking lot was too small, or five-stars bought by the owner.
SafeEats™ is just the facts. Inspector-issued scores. Violation codes. Official records. Did the health department pass this restaurant or not?
Trust & Quality Controls
The safeguards that make SafeEats™ a defensible, auditable dataset.
Source provenance
Every record stores its source jurisdiction, the original native score, and a freshness timestamp. Any grade can be traced back to its issuing authority — no anonymous data.
Documented normalization
Each jurisdiction's conversion logic is published on this page — no black-box scoring. The mapping from local violations to a 0–100 score is deterministic and auditable.
Freshness tracking
Live-API results are queried at request time. Cached records carry a cached_at and last_api_check timestamp so staleness is visible, not hidden.
Relevance filtering
Post-fetch name-relevance filtering strips false matches from permissive government search APIs — so 'Chipotle' returns Chipotle, not adjacent records the API bundled in.
Mandatory deduplication
Every result set is deduplicated by normalized name + address prefix before display, collapsing the same establishment that reappeared across API pages or sources.
Unverified-result filtering
AI results the model could not verify (confidence = 'none') are dropped entirely, as are restaurants confirmed closed. 'Restaurant found, no inspection details' is kept and labeled — not invented into a score.
Verification sourcing
AI results include the URL where the restaurant was confirmed (health dept site, official listing). If it can't be verified, it's not shown — zero hallucinated records.
Geo-fencing
Location validation runs on raw AI output before any field override. A result whose city/address doesn't match the requested region is rejected — so a Dubai search never returns a Miami strip mall.
No review contamination
SafeEats™ never uses consumer reviews (Yelp, Google Reviews) as a safety signal. Scores are inspector-issued only — opinions never move a grade.
Honest uncertainty
AI-estimated scores are always labeled. A 'U' (Unknown) grade means no records were found — never a fabricated score. We say so when we don't know.
Tiered model accuracy
Web-search results use Gemini 3 Flash for fast, real-time retrieval from official sources; preliminary results use GPT-5 Mini for instant restaurant identification. The right model for each step — never a single weak model for everything.
A Holistic Picture of Restaurant Safety
Most tools stop at a letter grade. SafeEats™ doesn't — because a single score can't tell you whether the water is clean, whether the kitchen has a history of repeat violations, or whether your family can even get through the front door.
Inspection History
Not just the latest grade — SafeEats™ shows the full timeline. Repeated violations, clean streaks, and trending patterns tell you whether a restaurant is consistently safe or just had one good day.
Water Quality
Even a spotless kitchen can't fix contaminated water. SafeEats™ pulls EPA water system data for the restaurant's area so you know what's coming out of the tap — contaminants, violations, and all.
Accessibility
Can your grandmother get through the door? Does it have an accessible restroom? A safety grade is meaningless if you can't enter the building. SafeEats™ checks ADA compliance for every US restaurant.
Source Transparency
Every score links back to the government database it came from. No anonymous data, no review-site opinions — just inspector-issued records you can verify yourself with one click.
Pest trackers tell you where the rats are. SafeEats™ tells you whether the restaurant is safe — for everyone, in every way that matters.
ADA Accessibility — A Personal Commitment
US RestaurantsAccessibility is deeply personal to our family. Knowing whether a restaurant is wheelchair accessible, has accessible restrooms, or adequate parking shouldn't require a phone call or a gamble — it should be right there alongside the health score before you ever leave the house.
What SafeEats™ Checks for Every US Restaurant
Accessible Entrance
Ramp or level entry confirmed
Accessible Restroom
ADA-compliant facilities on-site
Accessible Parking
Designated accessible spaces nearby
Automatic Doors
Power-assisted or automatic entry doors
📧 See wrong ADA info? Tell us.
If SafeEats™ displays incorrect accessibility information for any restaurant, use the Feedback page to report it. Daniel personally follows up with businesses where accessibility data is inaccurate — because every family deserves to know before they show up.
Tap Water Quality — Should You Order From the Tap?
US RestaurantsThe water your restaurant uses comes from a municipal supply — the same source feeding every soda fountain, ice machine, and pasta pot on the premises. For every US restaurant, SafeEats™ retrieves live EPA Safe Drinking Water Information System (SDWIS) data via a backend function — health-based violations, contaminant counts, and unresolved violations from the last 5 years — and computes a four-tier water quality grade (Excellent / Good / Drinkable / Not Recommended) directly on the restaurant detail card. When no public water system is on file for a location (e.g., private well areas), SafeEats™ falls back to a direct link to the Environmental Working Group (EWG) Tap Water Database so you can look up your zip code's report manually.
What EWG's Tap Water Database Covers
Lead & Copper
Neurological damage, especially in children
Nitrates
Dangerous for infants under 6 months
Bacteria (E. coli)
Gastrointestinal illness
Arsenic
Cancer risk with long-term exposure
Disinfection Byproducts
Trihalomethanes, cancer risk
Radionuclides
Radon, uranium — cancer risk
🇺🇸 US Coverage — Powered by EWG
On every US restaurant detail card, SafeEats™ retrieves live EPA SDWIS data and computes a visible water quality grade. When EPA has no system on file, it falls back to a direct EWG link for that restaurant's zip code. EWG tests go beyond EPA legal limits — they compare contaminants against stricter health guidelines, so you get the full picture, not just the legal minimum. International tap water data is on our roadmap.
Built for the American in Japan
Scan Any LanguageImagine arriving at a Tokyo konbini or a Osaka izakaya — menus entirely in kanji, ingredient labels in fine-print kana, and a health inspection placard you can't read. As a parent, an allergy sufferer, or anyone navigating a foreign food system, this is genuinely stressful.
Real Pitfalls SafeEats™ Helps You Avoid
- 🦐Hidden shellfish — Japanese sauces often contain エキス (extract) from shrimp or crab. SafeEats™ flags these from the ingredient list even when buried in kanji.
- 📅Expiration labels — 賞味期限 (best before) vs 消費期限 (use by) look identical to untrained eyes. SafeEats™ translates both and explains the difference.
- 🥛"Domestic" labeling — 国産 means "Product of Japan" not "dairy-free." SafeEats™ reads country-of-origin and surfaces it clearly.
- 📊Nutrition per 100g — Japan doesn't use US-style "per serving" sizing. SafeEats™ converts and displays calories, fat, sodium in the format you understand.
- 🕌Halal & vegan flags — Japan's labeling standards differ from Western ones. SafeEats™ checks ingredient lists, not just packaging claims.
Point SafeEats™' camera at any sign, label, or menu in any language — it reads, translates, and surfaces what matters most in seconds.
Camera Scanner public beta: April 2026
UAE & Dubai — A World Leader in Food Safety
AI-ENHANCEDDubai Municipality is one of the most digitally advanced food safety authorities on the planet. Their open-data platform — Dubai Pulse — publishes food establishment data publicly. SafeEats™ uses AI web search to read official Dubai Municipality sources and present results on the same normalized A–F scale as New York, London, and Toronto. Results are clearly labeled as AI-estimated.
What SafeEats™ Covers for the UAE
Dubai Municipality
Dubai Pulse publishes food establishment data publicly. SafeEats™ uses AI web search (Gemini 3 Flash) to read official Dubai Municipality sources — results clearly labeled as AI-estimated.
Normalized A–F Grading
Dubai's risk matrix and compliance outcomes are normalized to SafeEats™' universal 0–100 score and A–F grade.
Halal Verification Flags
SafeEats™ surfaces halal certification status where published, critical for dining decisions across the UAE.
All Seven Emirates
Dubai, Abu Dhabi, Sharjah, Ajman, Ras Al Khaimah, Fujairah, Umm Al Quwain — covered via AI-assisted research.
🏆 Why Dubai Is a Model for the World
Dubai's approach to open food safety data is an international benchmark. By making inspection records publicly searchable, Dubai Municipality empowers both residents and the 20+ million annual visitors to make informed dining decisions. SafeEats™ uses AI to read these official sources and present them on the same screen as New York, London, and Toronto — clearly labeled as AI-estimated where no direct API connector exists yet.
Built on Esri's Public GIS Standards
Color scaleSafeEats™ adopts Esri's diverging health-risk color ramp — the same green→yellow→red visual language public health agencies use worldwide — so a safety score reads instantly at a glance, even for users with color vision deficiencies. This is an open/public GIS standard we build on; SafeEats™ is not formally partnered with Esri.
Learn more about Esri →Important Disclaimer
SafeEats™ is an informational research tool only. It is not affiliated with, endorsed by, or a substitute for any government health agency or official inspection authority.
Data accuracy is not guaranteed. Scores come from government open-data APIs or AI-assisted lookups of public records. Both may contain errors, omissions, or outdated entries.
AI-estimated scores carry additional uncertainty. For jurisdictions without a live government API, scores are estimated by AI reading publicly available sources — not official government records.
Always consult your local health department for authoritative records. By using SafeEats™ you acknowledge data may be incomplete and you are solely responsible for decisions made based on it.
About the Creator
Daniel Landeck
Global Citizen · University of Washington Graduate
Daniel built SafeEats™ as a genuine public service — transparent food safety data, free, because the public deserves better than a wall of government PDFs. As a doting father, knowing what's in the food his family eats isn't optional — it's personal. That same drive extends to anyone navigating foreign food systems, managing allergies, or simply trying to make an informed choice about where to eat tonight.
Built with for food safety transparency