Digital Menus & Technology
What "automated" actually means, how AI-powered allergen detection works, and why the data pipeline is the difference between compliance and liability.
"Automated allergen tagging" is one of the most over-used and under-defined phrases in restaurant technology. Some platforms use it to describe centralized data entry, where staff input allergens by hand into a single system. Others use it for basic keyword matching. And a smaller number mean what the phrase actually implies: AI that ingests ingredient lists, identifies allergens automatically, and updates the data in real time across every menu and every channel.
The difference matters. California's SB 68 takes effect on July 1, 2026, and it applies to physical menus, digital menus, kiosks, websites, and every ordering channel a consumer can touch. The allergen data displayed to guests has to be accurate, consistent, and current — or the operator carries the compliance and liability exposure.
This guide walks through what digital menu allergen tagging actually is, how AI-powered detection works, the spectrum from manual to fully automated, and how to evaluate platforms against the standard SB 68 now demands.
Digital menu allergen tagging is the identification and display of major food allergens on a digital menu item, so that consumers can see which items contain which allergens before they order. At its simplest, this means a menu item description is paired with a clear, consistent statement of the allergens contained in that item.
The mechanics of how allergen tagging happens vary dramatically from platform to platform. Some systems rely on staff to manually tag each menu item. Some use centralized databases where a corporate team enters allergen data once and pushes it to all locations. And some use AI to analyze ingredient lists and automatically identify allergens from the ingredient data itself.
Accuracy matters more here than in almost any other area of restaurant technology. A mistagged allergen on a menu is not a typo — it is a potential medical emergency for a guest with a serious food allergy, and a liability event for the operator. Under SB 68, it is also now a regulated compliance failure. The platform you choose is a decision with legal, operational, and safety consequences.
True AI-powered allergen tagging is a pipeline, not a single step. It starts with ingredient data and ends with a human expert verifying the result. Here is how the process actually works:
AI is only as good as the data powering it. This is the part most discussions of "automated" tagging gloss over. For the output to be reliable, the input has to be accurate — which means pulling data from authoritative sources rather than relying on whatever description is on the menu.
In practice, this means: accurate menu data pulled directly from the restaurant's menu management provider, accurate recipes pulled from back-of-house inventory management systems, and exact product specifications pulled from all major food suppliers and distributors. When the underlying data is clean, the AI has something real to analyze. When the data is fragmented, outdated, or self-reported by memory, the output is only as reliable as the weakest input.
The AI parses ingredient lists using natural language processing. This is more complex than keyword matching. An ingredient called "natural flavors" might contain dairy derivatives. "Hydrolyzed vegetable protein" might be wheat-based. "Lecithin" is often soy-derived. The AI cross-references ingredients and their sub-ingredients against comprehensive allergen databases to identify not just the obvious allergens, but the hidden ones buried several layers deep in a supplier's formulation.
Not every ingredient maps cleanly. Some have ambiguous sourcing. Some are shared across multiple allergen categories. Some come from suppliers with incomplete documentation. A well-designed AI system assigns a confidence score to each allergen tag and flags anything below a threshold for human review. Edge cases like cross-contamination risks, shared production facilities, and regional supplier substitutions all get surfaced rather than silently guessed.
This is the step most operators assume does not exist — and it is the step that matters most. Food allergens can be a matter of life and death. While AI dramatically streamlines the tagging process and improves accuracy at scale, having a qualified human expert as the final reviewer is non-negotiable for a compliance-grade system.
Foodini keeps a team of registered dietitians in the loop as the final layer of quality assurance. Every AI-generated allergen profile is reviewed by a credentialed expert before it is published to a live menu. Anything the AI flags for review, anything with a confidence score below threshold, and anything that looks unusual is verified by a dietitian. This is also how Foodini catches errors in the source data itself — and in practice, errors in client-provided data come up often enough that the QA step routinely surfaces problems operators did not know they had.
Platforms that market themselves as "automated" sit along a spectrum. Understanding where a given system actually falls on that spectrum is the single most important evaluation step an operator can do.
Staff enter allergens into a system one menu item at a time. This is the starting point for most chains and the most error-prone approach. It does not scale past a handful of locations, cannot keep up with supplier changes, and depends entirely on the knowledge of whoever is entering the data. A kitchen manager who does not realize that Worcestershire sauce contains anchovies will not tag fish as an allergen on any item that uses it.
A step up from manual, this approach centralizes the allergen data in a single database. Corporate staff enter allergen information once, and the data propagates across locations. This reduces the version control problems of pure manual tagging, but the actual allergen identification is still happening by hand. If the person entering the data misses a hidden allergen, the error propagates to every location instantly. "Centralized" is not the same as "automated."
The highest tier on the spectrum. AI ingests ingredient data from authoritative sources, identifies allergens automatically using natural language processing and allergen databases, flags edge cases for review, and routes everything through a final quality assurance layer staffed by registered dietitians. The operator is not relying on a human to manually identify every allergen on every item — they are relying on a system where human expertise is applied where it matters most: verifying the output before it reaches consumers.
All three approaches can be described as "automated" in marketing copy. Only the third actually removes the primary source of allergen tagging errors: human identification of allergens across thousands of ingredients. For a 20-location chain under SB 68, the difference between centralized manual entry and true AI-powered detection is the difference between a scalable compliance system and a scalable error system.
SB 68 requires disclosure of nine major food allergens. But compliance in a single jurisdiction is a narrow frame for a platform decision that should serve the operator for years. Foodini's tagging goes well beyond the nine SB 68 allergens, covering more than 150 different allergens, intolerances, and dietary needs.
This breadth has direct practical value. It ensures compliance not just with SB 68, but with other jurisdictions as allergen disclosure laws expand. Sesame was added as the 9th major allergen in the US recently, and it is likely that additional allergens will be added to the mandatory list over the coming years. Canada requires disclosure of 11 mandatory allergens. The European Union requires 14. Operators with international exposure — or operators expecting the US list to grow, which it almost certainly will — need a platform that can handle the broader set without a system overhaul.
It also covers lifestyle and preference diets, which are increasingly important to consumers and increasingly important to restaurant revenue. Vegan, vegetarian, halal, kosher, keto, low-FODMAP, gluten-free by preference, and dozens of other dietary categories matter to a meaningful share of diners — and tagging them alongside allergens turns a compliance requirement into a revenue driver. A guest who can filter a digital menu by their specific dietary needs is a guest who is more likely to order, more likely to spend more, and more likely to come back.
Once the tagging data is in place, the implementation question is where that data needs to show up. Under SB 68, the answer is: every digital surface a consumer can touch. Foodini's technology powers allergen-flagged menus anywhere a menu exists:
The critical point is that all of these surfaces must display the same data at the same time. A guest who checks the allergens on a restaurant's website should see the exact same information as a guest ordering through a kiosk in the same store. A change to a recipe or supplier should propagate to every surface simultaneously — not several days later, and not with one surface updated while another still shows yesterday's data.
Testing and validation before go-live are essential. Every location, every format, every channel. A chain-wide rollout that leaves one ordering channel running stale allergen data is a rollout that created a new compliance gap instead of closing one.
If you are comparing platforms for SB 68 compliance, here are the questions that actually matter:
The language of "automated allergen tagging" covers a wide range of very different approaches. For an operator facing a July 1, 2026 compliance deadline — and the legal and liability exposure that comes with it — the specifics of how a platform actually identifies, verifies, and propagates allergen data are the decision. Marketing terms are not.
Foodini's approach combines AI-powered detection with authoritative data sources and registered dietitian QA. The AI does the heavy lifting at scale. The human experts ensure the output is accurate enough to stand behind. And the data propagates in real time to every digital surface where a menu lives — which, under SB 68, is every one of them.
Recently named one of Fast Company's Most Innovative Companies in restaurants, dining, and food services for 2026, Foodini is the dietary intelligence platform built for exactly this challenge.
Foodini automates allergen disclosure across every menu, every location, and every digital channel — with registered dietitian QA built in.
See how it works for your operation →