← Sticker Wall AGENT · DETAIL

Restaurant QR Code AI Sommelier: Dish-and-Wine Pairing Driving Monthly Revenue of 12,000 RMB

Workflow: Customers scan the QR code on the table to order food. The system reads the dishes already ordered for the current table

AGENT

Key Fields

FIELD STAMPS

IndustryE-commerce / Retail

RegionChina（国内）

ScaleSME

ChannelOnline

🔧 Workflow

Customers scan the QR code on the table to order food. The system reads the dishes already ordered for the current table, and based on dish flavor tags, cooking methods, and real-time wine inventory, calls a large language model to generate 1 to 3 wine-pairing recommendations, complete with layman-friendly tasting notes and serving temperature suggestions. Customers add wine directly to their cart via the mini-program. Upon order completion, revenue is shared with the restaurant based on an agreed-upon ratio. At the same time, records of every recommendation, click, add-to-cart, and transaction are logged to weekly optimize tag mapping and recommendation scripts.

🛠 Setup Requirements

Requires building a WeChat Mini Program or H5 page, integrating the wine list database, dish tag table, and wine inventory/sales/purchase spreadsheet. The backend calls the large language model API to generate recommendation content, and dynamic QR codes are bound to table numbers. The technical threshold is moderate, requiring proficiency in React or Vue plus a Serverless backend to get started, with cloud functions handling recommendation requests and order callbacks. Initially, spreadsheets can be used to import wine lists and dishes, and a test version can be launched within two weeks, followed by the gradual integration of restaurant POS or QR code ordering systems.

🧰 Toolchain

🔧 WeChat Mini Program
🔧 Large Language Model API
🔧 Serverless Cloud Functions
🔧 Wine Inventory and Sales Spreadsheet
🔧 Dynamic QR Code Generation Tool
🔧 Data Dashboard (for conversion rate statistics)

💰 Revenue

Single-store wine increment is about 3,000 to 5,000 RMB. Calculated at a 20% commission, monthly revenue is 600 to 1,000 RMB; covering 12 stable stores yields a monthly revenue of about 12,000 RMB. If concurrently connected to wine supplier rebates, an additional 3% to 8% channel commission can be extracted per bottle of wine, raising total revenue to around 15,000 RMB. Core revenue comes from wine add-on commission sharing, and secondary revenue from supplier rebates, which require real transaction data as the basis for sharing.

💸 Cost

Large language model API invocation costs are about 50 to 200 RMB per month, WeChat Mini Program certification and Serverless cloud function hosting are about 100 to 300 RMB per month, totaling no more than 500 RMB. The biggest cost is the human time and effort for early offline promotion, wine list entry, and inventory verification, as well as profit concessions given to restaurants for free trials during the initial promotion period.

⏱ Time Investment

In the early stage, invest 20 hours per week running stores, entering menus, maintaining inventory, and handling abnormal orders; after running smoothly, invest 5 hours per week updating menus, supplementing new wine selections, and analyzing conversion data, with some inventory partially synchronized automatically via web scrapers or spreadsheets.

🚀 Getting Started

First, find 3 to 5 local Chinese restaurants, bistros, or day-to-café-night-bars that have QR code ordering but lack professional sommeliers. Offer a two-week free trial, and use actual wine increment data to convince the owner to sign a revenue-sharing agreement. Second, map high-frequency wines with dish flavors to form reusable recommendation templates, and bind the wine supplier rebate pipeline.

🔑 Keys to Success

✅ Bind real-time wine inventory and revenue-sharing pipelines to avoid recommending out-of-stock wines
✅ Recommendation reasons must be explainable, using layman terms such as taste profile, temperature, and grease-cutting properties to make customers willing to order
✅ Offer free trials first to secure real incremental data before negotiating long-term revenue sharing
✅ Prioritize high-margin, steadily circulating wines to reduce restaurants' resistance to changing suppliers
✅ Consolidate every recommendation and transaction data to continuously optimize tags and scripts

⚠️ 风险

⚠️ Restaurants update wine lists frequently; inaccurate inventory can lead to recommendation failures or customer complaints
⚠️ Some restaurants already have fixed wine suppliers and are unwilling to change revenue-sharing partners or open interfaces
⚠️ Customers have low trust in QR code recommendations, and early conversion rates may fall short of expectations
⚠️ The large language model may generate exaggerated or non-compliant wine descriptions, requiring manual review of boundaries

📌 Real Cases

📌 Japanese sake sales support agencies provide sake recommendations via in-store QR codes, proving that the QR code wine-pairing model can work in real catering scenarios and collect user preferences
📌 Domestic wine private-domain distribution platforms such as Jiulihang use mini-programs for secondary distribution and cloud warehouse fulfillment, verifying the feasibility of wine-sharing pipelines and rebate mechanisms
📌 Bar dine-in catering systems already have complete source code for sale on platforms like Huzhan, indicating that the infrastructure for restaurant QR code ordering and wine add-ons has ready-made references

Sources