Common Questions

A scan starts by visiting the restaurant menu URL with a headless browser, which handles single-page menus that lazy-load their dish list. A parser extracts every dish, with its name, ingredients, and allergen tags wherever the restaurant publishes them. A language model then audits each dish against a vegan rule set built from years of menu reading - the dairy traps in sauces and breads, the swaps every chain knows how to do, the ingredients that have to be confirmed at the counter. The output is a per-item confidence score plus the exact wording to use at the counter so the kitchen can ship the dish correctly.

We publish a confidence score on every item so the uncertainty is visible. In aggregate, the high-confidence calls are right the vast majority of the time. The places the scan is least reliable are locally-owned restaurants without a structured online menu, regional chains that change their ingredient suppliers between markets, and anywhere the kitchen has reformulated a dish without updating the website. We pair every item with a check-before-you-order section that flags the questions to ask the counter so a diner who wants to be doubly careful has a checklist.

Vegan Recon is a vegan-eating tool, not a medical-allergen reference. Where the underlying restaurant publishes allergen data, we surface it on the item card; where the restaurant does not, we mark the gap as something to confirm at the counter. If you have a clinically diagnosed allergy, the canonical source is always the restaurant's own allergen documentation plus a conversation with the kitchen. We do not recommend relying on any third-party tool, including ours, as the sole source of allergen information.

Coverage grows two ways. First, we curate a queue of high-traffic chains by hand and run scans on those whether or not anyone has requested them yet. Second, anyone can submit a menu URL through the scan flow and Vegan Recon will run the pipeline on it; the result is added to the public catalogue so the next visitor gets the answer instantly. If a chain you visit often is missing, request it; the queue is heavily shaped by user requests.

Send us a correction. Email the address on the contact page with the restaurant name, the item name, and what the kitchen actually said. Every correction trains the next round of scans. We also re-run scans on a regular schedule so menus that have changed since the last scan get fresh results, but a direct correction is the fastest signal we get.

The free tier handles three scans a month and full access to the saved catalogue of every scan anyone has ever run, which by itself is a useful tool. Saving items to your personal vault is also free for any signed-in user. The paid Upgrade tier unlocks unlimited scans and removes ads entirely. There is no plan to put the basic scan behind a paywall; the work that costs money is the language-model fees on new scans, and the upgrade tier covers those fees plus modest server costs.

Each scan caches the result for a window of time so the next visitor gets the answer instantly without re-running the pipeline. After that window, the next visitor triggers a fresh scan and the cache updates. We also run periodic re-scans on the most-trafficked restaurants so the cache stays current even if no one has requested a fresh scan recently.

Yes, with the caveat that chain coverage is more reliable than local coverage. Chains tend to publish structured menus with consistent ingredient lists across locations, which makes the scan very accurate. Local restaurants vary - some have clean, structured menus and the scan works well; others have a JPEG-of-a-printed-menu situation where there is nothing for the parser to pull from. We are working on a graceful degradation path for those cases.

Yes. Sign in (account creation is free) and you can save items to a personal vault that follows you across devices. The saved-items list shows the current state of each item so you know if a previously-saved order has changed since you saved it.

Coverage today is strongest in the United States and improving in Canada, the UK, and Australia. The scan technology itself is language- and country-agnostic; the limit is whether we have the chain in our index and whether the restaurant publishes a structured online menu. If you submit a menu URL from anywhere, the scan will run and the result will be available to other visitors.

Either use the thumbs-down button on the item card to flag it (anonymous, no sign-in required) or email the contact address with details. Flagged items move into a review queue; sustained negative signal pulls an item out of the public-facing display while the next scan refreshes the data.

Recon is the act of scouting a place before committing to it. Eating vegan at a non-vegan restaurant is exactly that: scout the menu, identify the safe paths, decide whether to go in. The brand keeps the scout-and-plan framing because it matches the way most experienced vegans already navigate restaurants. The product is a faster version of the recon you would do in your head anyway.