Alibaba Cloud Software / SaaS China Translated from Chinese

Leyan Technology (乐言科技)

Customer Service Automation

Leyan Technology's intelligent customer-service robot had to route enormous amounts of dialog traffic for ecommerce brands, but the company's self-built Kafka environment was becoming expensive, hard to scale, and too inflexible for precision consumption. Because the same messaging backbone also served AI and data systems, instability or coarse-grained routing directly affected the performance and extensibility of its AI SaaS platform.

Read the original story Explore all use cases

LeYu Zhuren serves more than 20 million interactions per day

cross-border AIGC solution serves more than 4 million overseas stores

overall cost dropped 37%

the company works with more than 60,000 ecommerce customers

Reported impact

Business outcome

Leyan lowered infrastructure cost for its AI service backbone while making the platform more adaptable for high-volume ecommerce support and new business scenarios. The RocketMQ-based redesign reduced overall cost by 37%, preserved support for a service that handles more than 20 million interactions per day, and gave the company a more precise and extensible messaging layer for both domestic and cross-border AI commerce services.

Solution

Leyan Technology rebuilt the messaging core behind its AI customer-service platform on Alibaba Cloud's serverless RocketMQ service. The new architecture keeps dialog distribution as the central event fabric for the LeYu Zhuren ecommerce robot, but replaces self-operated Kafka clusters with a managed message layer that can handle traffic spikes, resource elasticity, and differentiated delivery rules with less manual work. Messages from ecommerce conversations and downstream AI or data services are routed through RocketMQ topics and tags so consumer applications can subscribe with stronger precision than the earlier pipe-like design allowed. This gives Leyan a more cloud-native control plane for strong-consistency or ordered workloads, lowers the burden of operating multiple clusters, and preserves a reliable backbone for NLP, knowledge-graph, deep-learning, and large-model workloads that sit above the queue layer in its broader AI SaaS stack.

Data flow

Buyer requests from ecommerce channels enter Leyan's customer-service platform, are distributed through RocketMQ, and then consumed by AI dialog, recommendation, and data services. The processed outputs flow back into customer-service conversations or related downstream systems, while tags and filtering rules determine which business components consume which message streams.

Solution architecture

3 components · 3 layers

Compute
- Leyan GPT Extends the platform with large-model capabilities for industry solutions and new AI workflows.
Serving
- LeYu Zhuren Uses the queue backbone to receive customer requests and deliver AI-driven ecommerce customer-service responses.
Orchestration
- Message Queue for Apache RocketMQ Serverless Acts as the elastic event backbone for dialog distribution and downstream AI-service processing.

Architecture clues

Dialog-message distribution is treated as the central event fabric for the AI customer-service robot.
RocketMQ is used for stronger precision consumption with tags and SQL-style filtering instead of a coarse message-pipe model.
The platform evolved from self-built Kafka clusters to a managed serverless queue architecture to handle variable traffic more efficiently.
The same backbone supports both customer-service traffic and AI or data-system integration workloads.

Evidence from the source

For business-decoupling and strong-consistency scenarios, Leyan adopted Alibaba Cloud RocketMQ Serverless to deliver higher reliability and deterministic behavior. (original: 针对侧重于业务解耦、涉及较多后置逻辑处理的场景（如强一致性、顺序消息等），采用阿里云消息队列 RocketMQ 版 Serverless 系列，以满足高可靠性与确定性需求。)

Kafka also acted as the data channel among AI and big-data systems, so ensuring smooth messaging was critical to the broader platform. (original: Apache Kafka 还作为各数据系统（如 AI、大数据等）之间的数据通道。因此，确保其消息服务流程的顺畅至关重要。)

Leyan's core business includes the LeYu Zhuren intelligent customer-service robot, which uses NLP, knowledge graphs, and deep learning to handle buyer consultations, issue resolution, recommendations, and relationship maintenance. (original: 其自主研发的电商智能客服机器人“乐语助人”... 基于自然语言处理、知识图谱、深度学习等人工智能技术，具备充分的语言理解能力，可以模拟金牌客服的回复逻辑，执行买家咨询接待、业务问题处理、智能推荐、客情维系等任务。)

The new architecture cut cost by 37% while supporting an AI SaaS platform that already serves more than 20 million people a day and over 60,000 ecommerce customers. (original: 云消息队列助力乐言科技降本 37%... “乐语助人”每天服务超过 2000 万人次，与六万余家电商客户合作。)

Transparency Evidence & methodology 95/100 reference value · 95% confidence · 1 source

Review how this reference was scored and trace each normalized claim back to its documented source.

Scoring methodology

How scores are calculated

Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.

Customer reference value95/100

Problem clarity20/20
Measures whether the customer's operational problem is specific enough to compare.
Implementation depth25/25
Rewards named components, data flow, integrations, and architecture detail.
Outcome specificity20/20
Measures whether the reported benefit is concrete rather than a generic claim.
Quantified impact25/25
2 metrics describe measurable impact.
Deployment scale5/10
Rewards evidence of a live or scaled deployment and meaningful integration scope.

Evidence confidence95%

Source authority20/25
The main evidence is an official vendor-published customer story.
Claim traceability25/25
12 of 12 key fields link to a named source.
Direct evidence20/20
4 direct source quotes retained.
Date provenance10/10
The publication date comes from the source page or publisher.
Source transparency10/10
1 source reference document publisher, URL, authority, retrieval, and supported fields.
Implementation specificity10/10
Rewards source-backed components, data flow, quantified impact, and a specific outcome.

Traceability

Sources & provenance

Field coverage shows which facts each source supports. Dates are shown only when documented by the source.

Alibaba Cloud
云原生加速电商行业赋能，云消息队列助力乐言科技降本37%_互联网与信息技术_阿里云客户案例

First-party
https://www.aliyun.com/customer-stories/internet-and-info-tech-2025-leyantech

Source type
Vendor-published customer story
Supports
- Customer
- Region
- Company size
- Business problem
- Solution
- Products
- Solution components
- Data flow
- Integrations
- technical area
- use case category
- Business outcome
- outcome category
- Metrics
- Architecture
- evidence quotes

Date evidence

Published on source page Apr 30, 2025 · Alibaba Cloud · First-party
Open date source

Not separately sourced Industry. Treat these normalized attributes as lower-confidence until a dedicated source is added.

Similar use cases

See all like this

Alibaba Cloud Archived · Jul 21, 2026 EaseMate Software / SaaS United States EaseMate's AI image and video generation required massive computational power with low-latency delivery. Content safety became paramount — with millions of AI-generated assets monthly, the company needed robust… Customer Service AutomationReal-time Analytics Alibaba Cloud Model Studio · Qwen

Alibaba Cloud Archived · Jun 9, 2026 Akool Software / SaaS United States Akool identified a major gap in the B2B AI video market, where existing tools lacked the stability and usability required for real-world business use cases. Large companies needed solutions that were both powerful and… AI-powered video translation into over 180 languages and dialects in minutes, a process that traditionally took… Customer Service AutomationReal-time Analytics Alibaba Cloud Model Studio · Qwen

Alibaba Cloud Dec 19, 2025 Yuyi Technology (语忆科技) Software / SaaS China Ecommerce brands in the maternal-and-infant segment were struggling with scattered consumer data, slow service response, weak demand mining, and delayed discovery of after-sales issues hidden in refunds and… frontline customer-service conversion improved 30.8% Customer Service AutomationMarketing Personalization LabelPaas · NeoSight 5.0 · NeoTrainer AI