A support desk where AI agents answer in WeChat, WhatsApp, and Slack
Customer questions arrived in WeChat, WeCom, WhatsApp groups, and email, and every answer meant a person checking the warehouse and shipping systems by hand. Every message now becomes a ticket, AI agents resolve the routine ones from live system data, and the team steps in only when a case needs judgment.
Client
Cross-border logistics provider
Industry
Cross-Border Logistics
Timeline
Ticketing and one channel first, then agents and the remaining channels
Key takeaway
Customers keep asking where they already ask. The answer comes back in the same thread, from live data, usually before anyone on the team has to look.
This provider served shippers who did not want another portal. They asked in the channels they already lived in — WeChat and WeCom on one side, WhatsApp groups on the other, email for the rest — and expected an answer there. Behind each question was the same routine: find the order, check the warehouse system, check the carrier, check the inbox, write the reply, and hope nothing was missed in a group chat of forty people. We built a CRM that sits on every channel and every operational system, turns each message into a ticket, and puts AI agents on the first pass. Agents answer what the data can answer, hand the rest to the right person with the context already gathered, and post the reply back into the conversation it came from.
The challenge
Support was a person reading chat groups. Questions were buried between messages, nobody could say which ones were still open, and the same “where is my shipment” question got answered three times or not at all. Each answer meant logging into the warehouse system, the carrier portal, and the inbox. As customers and groups grew, the team was spending its day retrieving facts instead of solving the problems that actually needed them.
Questions lost in the chat
Requests arrived across WeChat, WeCom, WhatsApp groups, and email, with no ticket, no owner, and no way to know what was still waiting.
Every answer was a lookup
Order status, stock, and delivery facts lived in the warehouse system, carrier tools, and email, so the simplest reply started with three logins.
Headcount was the only lever
More customers meant more groups and more messages, and the only way to keep response times down was to put more people on chat.
What we built
Every inbound message, in any channel, opens or updates a ticket in one CRM. AI agents read the request, pull the answer from the warehouse, shipping, and email systems, and resolve it when the data is clear. When a case needs a decision — a claim, an exception, an unhappy customer — it goes to the support team with the history and the facts already attached. The team reviews, adjusts, or approves, and the reply goes back into the same WeChat, WeCom, WhatsApp, or Slack thread the customer used.
- One CRM for every conversation across WeChat, WeCom, WhatsApp groups, email, and Slack
- Automatic ticket creation from any message, with owner, status, and history
- AI agents that resolve routine requests using live warehouse, shipping, and email data
- Escalation to the support team, with the context gathered before a person opens the ticket
- Human review and approval on replies that need judgment before they are sent
- Replies posted back into the original thread, so customers never change channels
Before & after
Before
- Questions sat in WeChat, WhatsApp, and email with no ticket or owner
- Every answer started with checking the warehouse, carrier, and inbox by hand
- Growing chat volume meant putting more people on the groups
After
- Every message becomes a ticket with a status and an owner
- AI agents answer routine requests from live system data
- The team handles exceptions, and replies land back in the original thread
Built with
Our approach
01
Audit
We read through real conversations across every channel and sorted which questions the systems could answer and which needed a person.
02
Build
We connected the channels and operational systems to one CRM, then put agents on the requests with a clear, data-backed answer.
03
Pilot
Agents drafted replies for team review first. Once the answers held up, routine tickets were resolved automatically.
04
Support
We tune the agents, add channels like Slack, and widen what they can resolve as the team trusts more of the queue to them.
Frequently asked questions
More engagements