TeleAgent Crosses 1.5 Million Users: China's AI Agents Are Learning to Get Things Done

Modern office worker using AI software on computer

Team of business professionals collaborating in modern office

Shanghai skyline with digital network connections

Human hand touching glowing AI interface screen

Artificial intelligence has spent years getting better at understanding us. Now it is learning to work for us. At the opening of PT Expo China 2026 on October 9, 2026, China Telecom announced that its workplace AI agent, TeleAgent, has passed 1.5 million registered users with more than 200,000 daily active users — a remarkable jump from just 27,000 registered users in mid-June.

The milestone was announced by Liu Guiqing, Executive Director, President and Chief Operating Officer of China Telecommunications Corporation. His framing was blunt and memorable: "TeleAgent marks the shift of AI from understanding what is said to getting things done." It is a line worth taking seriously, because the growth curve behind it suggests this is not just another chatbot demo.

From 27,000 to 1.5 Million in Months

TeleAgent was first unveiled at the World Artificial Intelligence Conference (WAIC) 2026 in the summer. In the few months since, its user base has grown more than fifty-fold. That kind of acceleration is rare even in China's fast-moving AI scene, and it tells us something important: businesses and professionals are not just curious about AI agents — they are putting them to work.

China Telecom reads the milestone as evidence that autonomous AI agents capable of end-to-end task execution are moving from conceptual showcases into large-scale enterprise adoption. In other words, the age of the AI demo is giving way to the age of the AI employee.

What TeleAgent Actually Does

TeleAgent is a native AI application for the workplace, built by China Telecom's AI arm, TeleAI. Instead of chatting back and forth, you drop your files into its workspace — drag and drop is enough — and the agent carries the task through to completion on its own. It serves individual professionals, small and medium-sized enterprises, and large state-owned conglomerates alike.

Two editions show how broadly it is being aimed:

  • Enterprise SaaS edition: supports multi-device collaboration and scenario customization, so whole teams can plug the agent into their existing workflows.
  • Private-deployment edition: released in mid-August, it keeps models and data entirely inside the customer's own local environment — a crucial option for companies and institutions that cannot let sensitive data leave their walls.

It runs natively on Windows, macOS, and leading Chinese enterprise operating systems including Kylin OS and UnionTech OS (UOS), and individual users need no complex server setup to get started. The message is clear: this is built for the real office, not the research lab.

The Skill Marketplace Explosion

Perhaps the most striking number in the announcement is not the user count but the skill count. TeleAgent's skill marketplace has grown from roughly 40,000 skills to more than 50,000 skills — and 99% of them were built directly by end users solving real productivity problems.

That statistic reveals the real engine behind TeleAgent's growth. Users are not waiting for China Telecom to ship features; they are building small task-solving "skills" themselves and sharing them with everyone else. This kind of user-generated ecosystem has a powerful flywheel effect:

  • More users create more skills
  • More skills make the agent useful for more tasks
  • More useful agents attract even more users

It is the same dynamic that made app stores dominant — except here the "apps" are AI capabilities anyone can build without writing traditional code. For anyone watching the global AI race, this is a template worth studying: the winning agent platform may not be the one with the smartest model, but the one with the most contributors.

Why This Matters Beyond China

TeleAgent is part of a larger Chinese strategy. China Telecom has gone all-in on AI as a business model — it became the world's first telecom operator to sell AI tokens like a utility, with monthly packages starting from RMB 9.9. Its TeleAgent office assistant reportedly consumes more than 200 billion tokens a day, and the operator has aggregated 170 large models onto its token management hub while building 420 industry-specific agents for verticals including government, transport, and manufacturing.

For readers in Pakistan and around the world, three takeaways matter:

  • The agent shift is real. When 1.5 million people register for an AI that does tasks rather than answers questions, the industry's center of gravity is moving.
  • Data sovereignty sells. The private-deployment edition exists because enterprises — including state-owned giants — demand control. Expect every serious agent platform to offer a private version soon.
  • Ecosystems beat features. A 50,000-skill marketplace built almost entirely by users is a moat no single feature release can cross.

The contest of the next decade may not be about which AI model is the smartest. It may be about which one actually shows up to work.

Disclaimer: user figures and product details reflect China Telecom's announcement and may change; pricing details mentioned are illustrative and subject to change.

Post a Comment

Previous Post Next Post