Multilingual Menus — 23 Languages, AI-Translated, Pool-Cached

India has 22 official languages and at least 10 more that work conversationally across cities. For a restaurant in Bandra, the diners might speak English, Marathi, Hindi, Gujarati, Tamil, Konkani — sometimes in the same group, ordering the same dish for different reasons. A printed menu, locked to one language, ignores this reality. A QR menu is the only practical surface where multilingual support is actually achievable. QRSeva's multilingual feature is the platform's most under-rated competitive moat.

What multilingual menus do

When you enable a second language for your restaurant, QRSeva translates every text field on your menu — category names, item names, descriptions, dietary tag labels, even your welcome message — into the new language. The translation happens server-side, in the background, and finishes in under 30 seconds for a typical 50-item menu.

When a diner scans your QR, the menu page shows a small flag selector at the top right. They tap their preferred language and the entire menu instantly re-renders. Their choice is remembered on their device, so the next time they scan (or any QR menu on QRSeva), it opens in the language they previously selected.

The 23 languages we support

QRSeva covers the major Indian languages — English, Hindi, Marathi, Tamil, Bengali, Gujarati, Punjabi, Kannada, Malayalam, Telugu, Odia, Assamese, Urdu — plus key tourist languages: French, Spanish, German, Italian, Mandarin, Japanese, Korean, Arabic, Portuguese and Russian. The set was chosen by tracking which languages QRSeva diners actually selected over a six-month observation period; we add languages when usage data justifies them.

If your venue serves a language we don't yet cover, write to support@qrseva.com with diner-count data and we will prioritise it. The platform is built so adding a new language is a configuration change, not a code change.

How AI translation works on QRSeva

When you toggle a language on, the platform queues every menu field for translation. Each field passes through a language model that has been specifically tuned on Indian restaurant terminology. So "Paneer Tikka" becomes "पनीर टिक्का" in Hindi (transliteration, because the dish name is recognised as Indian) and "Pannier Tikka" — keeping the brand name intact — in Spanish (because that's a meaningful name, not a translatable phrase).

The model preserves dish names in their canonical form across languages, while translating descriptions in full. "Crispy fried paneer marinated in yoghurt and spices" becomes a faithful Hindi translation, but the dish name "Paneer Tikka" stays. This balance — keep what should be preserved, translate what should be localised — is the hardest part of menu translation, and it's where generic translation tools fail.

Why pool-caching changes the economics

Most QR menu platforms charge per-translation. So if you have 50 items and you turn on Hindi, you pay 50 translation credits. QRSeva does this differently. The first restaurant to translate "Paneer Tikka" to Hindi triggers an AI call and stores the result in a network-wide pool. Every subsequent restaurant that has "Paneer Tikka" gets that translation for free — pool-cache hit.

In practice, this means a new restaurant in a Hindi-speaking market typically gets 70-85% of its menu translated for free. The 15-30% that's unique to their menu (regional specialties, custom names) triggers actual AI calls and consumes credits. The bigger the QRSeva network grows, the cheaper multilingual gets for every new restaurant.

This effect compounds. A restaurant joining QRSeva today benefits from translations made over the last two years. A restaurant joining two years from now will see almost zero AI cost on common dishes — most of their menu will already be in the cache.

How diners experience multilingual menus

When a diner scans your QR, the menu opens in their device's primary language — if you've enabled that language. If they prefer a different one, they tap the flag in the corner and pick from your enabled set. The choice persists across visits.

Crucially, the URL stays the same regardless of language. Your QR code doesn't change. The menu URL doesn't change. The diner just sees the content in their language. This means your QR works equally well for English, Hindi, Marathi and Tamil speakers — one printed code, every language served.

What gets translated, what doesn't

Translated automatically: category names, item names (except canonical dish names), item descriptions, dietary tag labels, allergen notes, welcome messages, button labels (Order, Add, etc.), template UI strings.

Not translated: prices (kept in your base currency — see the currency feature for separate currency display), brand names (your restaurant name, dish names that ARE brands), specific cuisine terms (Murgh, Paneer, Tikka, etc., kept in canonical form so customers can recognise them), proper nouns in general.

You can override any individual translation manually from the dashboard. The auto-translation gives you 95% there; the dashboard lets you polish the 5% where the AI was over-literal or where you want a brand-specific phrasing.

When to enable a second language

• You serve tourist neighbourhoods (Bandra, Connaught Place, Koregaon Park). Enable English + the dominant tourist language.
• You serve a city with strong regional identity (Pune, Chennai, Kolkata). Enable English + the regional language so locals feel at home.
• You serve a NRI / diaspora-heavy clientele. Enable English + Hindi (or whichever language the diaspora maintains).
• You serve corporate offices. Enable English. Most corporate dining defaults to English regardless of native language.
• You're in a tier-2 city with high regional-language usage and low English fluency. Lead with the regional language; English secondary.

Cost model

Adding a language to your restaurant is free. The AI translations that get triggered when you enable it (only for items not already in the pool cache) consume AI credits — typically ₹100-300 for a 50-item menu in a moderately well-covered language pair. After the initial translation, edits to individual items re-translate just that item, which is a few rupees of credit.

For languages with high pool-cache coverage (English, Hindi, Marathi, Tamil, Bengali, Gujarati), the initial cost can be as low as ₹50-100. For less-common languages with thinner caches, it's on the higher end of the range. Either way, the ROI is straightforward: every diner who can confidently read your menu is a diner more likely to order, which more than pays for the translation many times over.