All Case StudiesHospitality

The Menu Nobody Could Read

How AI bridges the language gap between Chinese restaurant kitchens and Kiwi customers -- turning 'What's Mapo Tofu?' into a $22 order instead of 'Just sweet and sour pork, thanks.'

4 min readUpdated 2026-04-03Based on Claude Sonnet 4 / GPT-4o

The Real Problem

Wei runs a Sichuan restaurant on Dominion Road. The food is outstanding -- hand-pulled noodles, proper mapo tofu with Sichuan peppercorn, dan dan noodles that regulars drive across Auckland for. His kitchen team trained in Chengdu. The flavours are authentic.

But 80% of his customers are Chinese. The Kiwi customers who do walk in scan the menu, don't recognise most of the dishes, and default to the safe choice: sweet and sour pork. It's the lowest-margin dish on the menu. They never try the water-boiled fish (NZD 32) or the cumin lamb ribs (NZD 28) because "Shui Zhu Yu" and "Zi Ran Yang Pai" mean nothing to them.

The menu has English translations, but they're literal and unhelpful. "Water boiled fish" sounds like plain poached fish, not the fiery, chilli-oil-drenched Sichuan classic it actually is. "Husband and wife lung slices" scares people off a dish that's essentially thinly sliced beef in chilli sauce.

This isn't just Wei's problem. NZ's Chinese restaurant sector is in crisis. Patronage has dropped over 30%, and 108 Chinese food businesses went into liquidation in the year to August 2025. A destructive price war has pushed set meal prices under NZD 10 across Auckland. Restaurants competing on price are racing to the bottom. The ones that survive will be the ones that can attract a broader customer base -- and that means reaching Kiwi diners who currently don't know what to order.

The language barrier extends beyond the menu. Wei's English Google reviews often go unanswered for months because he doesn't feel confident writing fluent English responses. Research shows 94% of customers read management responses to reviews before choosing a restaurant. Every unanswered English review is a missed chance to build trust with non-Chinese customers.

On social media, Wei posts in Chinese on WeChat (which has 180,000+ monthly active NZ users) and Xiaohongshu, but his Instagram and Facebook -- where Kiwi customers look -- are mostly empty.

Why Existing Tools Don't Solve This

In the US, a company called Tarro (formerly Voiceplant) serves Chinese restaurants with phone ordering and has grown to USD 85 million in annual revenue. No equivalent exists in New Zealand.

Google Translate can translate a menu word by word, but the results are often worse than no translation at all. "Mouth water chicken" (口水鸡) is meaningless in English. What a Kiwi customer needs to hear is: "Cold poached chicken in a numbing chilli-sesame sauce -- our most popular starter."

Generic review response tools don't understand the cultural context. A Kiwi customer leaving a review saying "the dishes were a bit different from what I expected" needs a response that acknowledges the unfamiliarity and invites them to try a guided experience next time -- not a standard "Thank you for your feedback."

No tool in the NZ market bridges the cultural and language gap that sits between Chinese restaurant kitchens and mainstream Kiwi diners.

How AI Solves This

Wei's OpenClaw AI assistant lives on the restaurant's website and WhatsApp. It speaks fluent English and Chinese, and it knows the menu inside out -- not just the names, but the stories, the flavours, and the best way to explain each dish to someone who's never tried Sichuan food.

Menu Discovery

A Kiwi customer browsing the website asks:

Customer: "What's good here? I've never had Sichuan food before."

AI: "Great question! If you're new to Sichuan, here's where I'd start:

  • Dan Dan Noodles (NZD 18) -- Think of it as a rich, savoury pork mince sauce over fresh noodles with a gentle chilli kick. It's our gateway dish -- almost everyone loves it.
  • Cumin Lamb Ribs (NZD 28) -- Crispy lamb ribs tossed with cumin, chilli flakes, and spring onion. Smoky, fragrant, and very moreish. Great to share.
  • Mapo Tofu (NZD 22) -- Silky tofu in a deeply savoury sauce with Sichuan peppercorn (that's the famous numbing tingle). It's bold but not blow-your-head-off spicy.

If you're nervous about heat, I can suggest the milder dishes. Or if you want the full Sichuan experience, I'll point you to the real firecrackers. What sounds good?"

That customer orders mapo tofu and cumin lamb ribs (NZD 50) instead of sweet and sour pork (NZD 16). The average spend just tripled.

Google Review Responses

When a new English review appears:

Review (3 stars): "Food was OK but we didn't really know what to order. Ended up getting sweet and sour pork which was fine but nothing special."

AI draft reply: "Thanks for coming in! Sweet and sour pork is a solid choice, but next time I'd love for you to try some of our Sichuan specialties -- the Dan Dan Noodles and Cumin Lamb Ribs are the dishes our regulars keep coming back for. If you're not sure where to start, just ask any of our staff or check out the 'New to Sichuan?' section on our website. We'd love to show you what Dominion Road Sichuan food is really about."

Bilingual Social Content

Every week, the AI generates paired content:

  • Instagram/Facebook (English): "The dish that converts first-timers. Our Dan Dan Noodles -- rich pork mince, hand-pulled noodles, and just enough Sichuan heat to make you order a second bowl. NZD 18. Open Tue-Sun from 11:30am."
  • WeChat/Xiaohongshu (Chinese): The same dish, but written for the Chinese diaspora audience with different hooks -- nostalgia, authenticity, hometown flavours.

Two audiences, two languages, one consistent posting schedule.

How We Set This Up

None of this works if the AI is just a standalone chatbot with no connection to your actual business. That's why BestAI builds a custom integration program -- a piece of software that bridges your AI assistant with the systems you already use.

For this kind of setup, that means:

  • Building a detailed bilingual knowledge base of every dish -- English descriptions that explain flavour, heat level, and texture in ways Kiwi customers understand, plus Chinese descriptions for diaspora audiences
  • Connecting the AI to your website as a live chat widget and to WhatsApp for direct customer conversations
  • Integrating with your Google Business Profile so the AI can draft English review responses for your approval
  • Setting up a bilingual social content pipeline that generates posts for both English (Instagram/Facebook) and Chinese (WeChat/Xiaohongshu) audiences

Here's our process:

  1. We map your current workflow -- We sit down with you and learn your menu, your story, and how you currently communicate with customers in both languages.
  2. We build the connections -- Our developers write a custom program (an API connector) that lets the AI talk to your website, messaging channels, and Google profile. No manual data entry, no copy-pasting between apps.
  3. We test end-to-end -- Every workflow gets tested with real scenarios before going live. We test with native English speakers to make sure the dish descriptions actually make people hungry, not confused.
  4. We maintain it -- When your menu changes seasonally, we update the knowledge base to match.

You don't need to be technical. We handle all the development -- you just tell us about your food, and we make the AI communicate it to everyone.

The Result

  • Higher average spend from Kiwi customers -- they order the authentic dishes (NZD 22-38) instead of defaulting to sweet and sour pork (NZD 16)
  • Every English Google review gets a thoughtful response -- building trust with the 94% of customers who read them
  • Dual-language social media presence -- consistent posts on both English and Chinese platforms without doubling the work
  • Broader customer base -- the restaurant stops being "that Chinese place Kiwis don't know how to order at"
  • Menu education at scale -- the AI explains dishes to hundreds of website visitors without pulling wait staff away from tables

What AI Can't Do Here

  • AI won't replace the warmth of a waiter personally recommending a dish tableside
  • AI can't fix food quality or consistency -- it only communicates what's already good
  • AI won't eliminate the language barrier entirely -- some customers still prefer to order in person
  • AI won't run your WeChat or Xiaohongshu accounts for you -- it drafts content, but you maintain the accounts
  • AI can't change the economics of the price war -- it helps you compete on experience instead of price

Who This Is For

  • Chinese restaurant owners whose English-speaking customer base is smaller than it should be
  • Restaurants with authentic, high-margin dishes that non-Chinese customers skip because they don't understand the menu
  • Any restaurant where English Google reviews go unanswered due to language confidence
  • Chinese food businesses that want to post on both English and Chinese social media without hiring two marketing people

Want This for Your Business?

Book a 45-minute workflow review and we'll show you exactly how this applies to your specific situation, no obligation, no fluff.

More Hospitality Case Studies

Replying to Google Reviews After a 14-Hour Shift

How AI drafts personalised review responses that reference specific dishes and feedback — so your Google presence stays active even when you're exhausted.

Read case study

Every Missed Call Is a Lost Booking

How AI answers your restaurant's phone during service — handling reservations, menu questions, and hours enquiries without pulling staff off the floor.

Read case study

Your Instagram Hasn't Been Updated Since Last Month's Special

How AI creates a weekly social media calendar from your daily specials, food photos, and events — keeping your feed alive without hiring a marketing team.

Read case study

One Video, Five Languages, Zero Translators

An AI tool that takes a single promo video and produces localized versions in English, Mandarin, Japanese, Korean, and Hindi with native voiceover and lip-sync.

Read case study

You Don't Know What Each Dish Actually Costs You

How AI turns supplier invoices into real-time food cost tracking — so you know your margins before they disappear.

Read case study

The Same 10 Questions, 50 Times a Week

How AI handles repetitive customer enquiries across WhatsApp, Facebook, and Google — so you stop typing 'Yes, we have gluten-free options' every day.

Read case study

A Professional Menu with AI Food Photography in Under an Hour

An AI tool that takes your menu items and generates a beautifully designed, multilingual menu with photorealistic food images for every dish.

Read case study

The Saturday Morning Catering Call You Never Answered

How AI captures corporate catering enquiries during your busiest rush -- so you stop losing $800 orders to the cafe down the road.

Read case study

Thirty Cents of Every Dollar Goes to Uber Eats

How one Auckland Chinese restaurant built its own ordering channel and kept the margin -- without learning to code.

Read case study

Raj and Priya Can't Answer the Phone During Dinner Rush

How AI takes every takeaway and catering call during your busiest hours -- so you stop losing orders while you're cooking.

Read case study

Fifteen Catering Enquiries Before Diwali. Three Replies Sent.

How AI qualifies catering leads, collects requirements, and prepares quotes -- so you never lose a wedding booking because you replied too late.

Read case study

Twenty Seats. Forty People Waiting. No System.

How AI turns a chaotic ramen shop queue into a smooth virtual waitlist -- so customers wait from the cafe next door instead of leaving.

Read case study

Forty Pieces of Salmon Nigiri in the Bin Every Night

How AI predicts exactly how much sushi to prepare each hour -- so you stop throwing profit in the bin at closing time.

Read case study

The Korean BBQ That Nearly Closed Its Dining Room

How AI handles bilingual bookings and guides first-time diners through Korean BBQ -- so you don't need to choose between answering phones and running service.

Read case study

Chimaek Without the Commission

How a Korean fried chicken shop built AI-powered direct ordering and stopped giving 30% of every delivery to Uber Eats.

Read case study

The Pho Shop That Stopped Feeding Uber

How a family-run Vietnamese restaurant built its own AI ordering channel and kept the 30% margin Uber Eats was taking.

Read case study

No Peanuts, No Problem

How AI gives every customer instant, accurate allergen information for every dish -- so your staff never have to guess about fish sauce or shrimp paste again.

Read case study

The Friday Night Phone Problem

How AI answers every call at a fish and chips shop during the Friday rush -- so two people can keep frying instead of one running to the phone.

Read case study

How an Independent Pizza Shop Fights Back Against Domino's

How AI phone ordering and smart upselling help a single-location pizza shop compete with chain-level technology -- without chain-level budgets.

Read case study

The Weekend That Books Itself

How AI handles bilingual hotpot reservations, manages the waitlist, and guides first-timers -- so your staff stay on the floor instead of the phone.

Read case study

The AYCE Profit Protector

How AI tracks consumption, predicts prep quantities, and protects margins at an all-you-can-eat BBQ restaurant -- without awkward waste penalties.

Read case study

The Monday Prawn Problem

How AI predicts exactly how much food to prepare for each buffet session -- so you stop throwing away $200 of prawns on quiet nights.

Read case study

The Tapioca Timing Problem

How AI tells your bubble tea shop exactly when to cook the next batch of pearls -- so you stop throwing away rubbery tapioca at closing time.

Read case study

One Photo, Four Platforms, Two Languages

How AI turns a single product photo into bilingual social media content for Instagram, TikTok, WeChat, and Xiaohongshu -- in minutes, not hours.

Read case study