Customer Service Software Guides & AI Agent Blogs | UdeskAgent on Duty | Serving Yili’s 10M+ Members in Seconds
article summary：This article introduces Udesk’s enterprise‑grade AI Agent deployed at Yili. It automates member phone‑number rebinding, unbinding, order and points queries via natural‑language‑driven API calls. The Agent cuts operation time to seconds, boosts agents’ daily throughput by 60% and lifts first‑contact resolution rate to 92%, improving both member experience and service efficiency for Yili’s massive member base.
Table of contents for this article
- 1. Behind Yili’s 10M+ Members: A Time-Consuming Bottleneck
- 2. In the Chat Window: An Agent Becomes a Capable Business Partner
- 3. Four Numbers: From Minutes to Seconds
- 4. A Faster Service Loop Feeds Brand Trust
- 》》Click to start your free trial of Omnichannel Systems, and experience the advantages firsthand.
Udesk's enterprise AI agents are built to get things done. Across retail, finance, manufacturing, government, and many other industries, they call APIs, execute transactions, and create tickets—embedding AI directly into real business workflows.
In this series, we profile AI agents already on the job at real enterprises. First up: Yili Group.
A support agent types a line in the chat window: “Help me switch member 138xxxx6688’s bound number to 139xxxx3322.” Three seconds later, the screen confirms: “Resolved. SMS notification sent.”
1. Behind Yili’s 10M+ Members: A Time-Consuming Bottleneck
Yili consistently ranks among the world’s top five dairy companies. Every day, over 10 million members make repeat purchases and check their points across its system—powered by more than a dozen backend platforms: member center, order center, marketing center, coupon system, and more.
Agents handle plenty of requests to change or unbind phone numbers every day. These aren’t complex, but they’re tedious. The traditional workflow: agents manually log into back-end systems, look up IDs, verify identities, update bindings, and trigger SMS notices—taking 3 to 5 minutes per request. While members wait on hold, agents toggle between systems, and the next call can’t get through.
What Yili’s service center needed was an assistant that could resolve issues on the spot.
2. In the Chat Window: An Agent Becomes a Capable Business Partner
The AI agent Udesk built for Yili is a doer, not just a chatbot. When an agent types a natural-language request in the chat window, the agent runs through four steps behind the scenes: identify the member, call the right business API, execute the change or unbind, then confirm completion and trigger the SMS.
No system switching for the agent. No need for the member to leave the conversation. Once done, the agent simply tells the seat: “Resolved.”
These capabilities are already live across Yili’s service center for phone number changes, unbinds, order lookups, and point adjustments.
3. Four Numbers: From Minutes to Seconds
After the agent went live, four metrics improved dramatically:
| Member number change (per request) | 3–5 min → 10 sec |
| Old-number unbind (per request) | 2–3 min → 5 sec |
| Daily requests handled per agent | +60% |
| First-contact resolution rate | 70% → 92% |
4. A Faster Service Loop Feeds Brand Trust
When this entry point gets faster:
- Members don’t have to leave the chat—a smoother experience.
- Agents move from “processors” back to “service providers”—focusing on genuinely complex issues.
- Every request and resolution is logged, feeding back into member operations.
- A single well-handled inquiry builds more brand trust than an ad campaign ever could.
Which AI employee should we profile next? Let us know in the comments.
》》Click to start your free trial of Omnichannel Systems, and experience the advantages firsthand.
The article is original by Udesk, and when reprinted, the source must be indicated：https://www.udeskglobal.com/blog/agent-on-duty-serving-yilis-10m-members-in-seconds.html
AI AgentAI chatbotai customer service
