.png "Global Insight")
For the past two years, most people have met artificial intelligence as a chatbot — something that answers questions. This week, a very different signal came from China. According to reports from Media OutReach and Malay Mail on October 9, China Telecom's office agent TeleAgent has passed 1.5 million registered users, with more than 200,000 people using it every day. In mid-June that registered base was only about 27,000. That is not gradual growth. That is a jump.
Why It Matters Now
Executive Liu Guiqing framed the milestone at PT Expo China 2026 as a shift "from understanding what is said to getting things done." TeleAgent is built to execute end-to-end work: drop in files, let the agent sort, summarise, draft, and file across devices. It runs on Windows, macOS, and Chinese systems like Kylin OS and UOS, and it comes in both a cloud SaaS edition and a private-deployment edition that keeps data inside a company's own walls — a detail that matters for banks, government-linked firms, and anyone nervous about sending documents to a public cloud.
The marketplace detail is telling. TeleAgent's skill library has grown from roughly 40,000 to more than 50,000 skills, and the company says 99% were built by end users solving their own problems. When users build the tools themselves, adoption stops being a top-down IT rollout and starts behaving like an app store.
What This Means for Everyone Else
This is the direction the whole industry is moving. OpenAI, Google, and others have all been racing to turn chatbots into agents that can book, buy, code, and operate software. TeleAgent's numbers suggest enterprises will pay for agents that finish tasks reliably and privately, not just generate nice paragraphs. For workers, the question shifts from "will AI answer me?" to "which of my repetitive workflows will I hand over first — and who checks the agent's work?"
Follow Global Insight for daily explainers on the ideas shaping technology.