Haidilao: AI customer service and AI store inspection

Haidilao, the Chinese hotpot chain, has moved from iPad ordering to AI customer service, an AI decision system and camera-based store inspection. It brought in AI customer service and ByteDance’s Volcano Engine large-model stack in 2023, rolled the AI decision system out to stores nationwide from 2024, and reports 2.16x faster model inference.

Summary

Item	Detail
Company	Haidilao International Holding Ltd.
Industry	Retail
Use cases	Customer experience, business intelligence, process optimization
Functions	Marketing, business intelligence, customer experience, process optimization
Reported result	2024 revenue RMB 42.755B (+3.1%), net profit about RMB 4.7B (+4.6%), core operating profit RMB 6.23B (+18.7%), 1.1 million+ diners a day (group-wide results; the source does not attribute them to AI)
Moments rating	S-tier quick win
AI use-case score	95

Why it matters

Haidilao’s brand rests on service and consistent store standards, so it applied AI at both ends: customer-facing service and back-of-house inspection. The large models arrived on top of more than a decade of digital groundwork.

Case details

Background

Haidilao introduced iPad ordering in 2011 and moved its core systems to the cloud in 2016. At the end of 2024 it ran 1,368 restaurants and served 415 million customer visits that year; Sina Tech reported in November 2025 that its stores worldwide had passed 1,400. With many stores and demanding standards, food safety checks have long been hard. At a 2021 restaurant brand event, executive director Song Qing said manual inspection needed large numbers of inspectors and auditors, was inefficient, varied with inspector skill and often missed items.

What they did

• Customer service: in 2023 Haidilao introduced AI customer service and a data middle platform, and adopted Volcano Engine’s data and large-model stack across the board. AI voice service responds “almost within seconds”.
• Decisions: from 2024 an AI decision system was rolled out to stores nationwide, and in 2025 the company built an intelligent middle platform covering all its brands.
• Inspection: a digital supervision system generates inspection reports automatically; kitchen cameras flag non-compliant staff actions and alert in real time; remote inspection shows live store and equipment data; food safety issues go straight to headquarters. Song Qing said at the time that AR remote inspection was still being explored.

Results and data

The only AI-specific figure in the sources is 2.16x faster model inference. No numbers are published for customer service response times, inspection coverage or recognition accuracy. At group level, 2024 revenue was RMB 42.755 billion (+3.1%), net profit about RMB 4.7 billion (+4.6%), core operating profit RMB 6.23 billion (+18.7%) and table turnover 4.1 times a day. These are overall business results, and the sources do not attribute them to AI.

What to take from it

• More than a decade of digital foundations (ordering, cloud, data platform) came before the large models, and that order is worth copying.
• Moving inspection from “send someone to check” to “cameras watching continuously, with issues reported straight up” suits chains with many stores and uniform standards.

Limits

Restaurant groups with few stores or loose standardization will struggle to spread the cost of cameras and middle platforms. The public material lacks data on the effect of any single AI tool, so group profit growth should not be treated as AI return when evaluating similar projects.

Source

Original source: Sina Finance · Haidilao AI report

Additional source: Beijing News: Song Qing on AR remote inspection and food safety

Additional source: Securities Times: Haidilao 2024 results

This is an industry research entry: a publicly documented third-party AI case, not a Moments client case. For projects Moments delivered itself, see client cases.