Listen to the article
Tencent says its AI assistant processes more than 100 trillion tokens a day, but the company’s usage and revenue figures do not yet show whether its new tools for cost control, security and agent management deliver consistent returns for businesses.
Tencent says WorkBuddy has surpassed 10 million daily active users and processes more than 100 trillion large language model tokens each day, making it China’s largest AI application by daily token use. The figures, disclosed in the company’s third-quarter results, are Tencent’s own claims rather than an independent industry ranking. Token volume indicates substantial activity, but does not by itself show how much revenue the product earns or whether it is profitable. [2]
The scale is striking, but the business picture needs careful reading. Sina Finance reported that Tencent put WorkBuddy’s revenue over a recent seven-day period at an annualised rate of more than US$500 million, over twice the comparable figure for Tencent Meeting. It also reported that WorkBuddy’s third-quarter revenue was more than five times its second-quarter total. The annualised figure is a run-rate calculation, not revenue earned over a full year, and Tencent has not separately disclosed a complete account of the product’s profitability. [6]
At the Tencent Global Digital Ecosystem Summit on 29 October, the company outlined four enterprise-focused additions: an open platform, security measures, an observability centre and WorkBuddy Managed Agents, or WMA. The announcements address a broader question facing organisations: how to govern AI agents, connect them to existing work and judge whether they are delivering value at scale. InfoQ’s summit coverage reported that Tencent said the service had reached 60 per cent of China’s top 100 companies and half of its top 500, figures that remain company-reported rather than independently audited. [4], [5]
One obstacle to wider adoption is knowing where the costs are going. WorkBuddy’s observability centre is intended to show token consumption and expenditure across teams and use cases, as well as the steps an agent takes while completing a task. China National Radio reported that H3C used this information to guide resource allocation and introduce token incentives; its account put employees’ daily WorkBuddy conversations at 1.3 million. [4]
The same report said Suning.com used execution traces to identify inefficient agent interactions, then adjusted prompts, models and tools. The company reported that token consumption fell by 10 per cent after those changes. That result illustrates how usage data might help target savings, but it is a reported customer outcome, not an independently audited measure or a guarantee of similar reductions elsewhere. [4]
Security is another condition for bringing agents into sensitive work. Tencent’s security materials describe controls covering permissions, execution and data, alongside options including trusted computing and enterprise deployment arrangements. In a separate technical account, Tencent says its TC-AICC trusted execution environment is designed so that plaintext is visible only on the client and within the protected environment. These are descriptions of the provider’s stated safeguards, not independent proof that every deployment is risk-free. [7]
A further challenge is preserving useful workflows beyond the individual employee who created them. Tencent Cloud’s documentation describes WMA as a hosted service for building and running enterprise agents, with orchestration, isolated sandboxes and ways to distribute agents to staff through WorkBuddy. It also lists management features such as tracing, evaluation, security auditing and cost controls. The documentation sets out intended capabilities; it does not independently verify the business impact of particular deployments. [3]
Tencent’s own use provides one illustration of how such tools might connect with established systems. China National Radio reported that the company’s Pan Shi Xiao O assistant links to more than 300 core ERP interfaces, while capabilities from Tencent Docs and Tencent Cloud Drive support millions of tasks each day. Those examples suggest how agents could make internal services available through conversational workflows, but they are company-related claims rather than independently assessed results. [4]
WorkBuddy’s usage figures and reported adoption point to rising interest in enterprise agents, but they leave practical questions for prospective customers. Organisations will need to assess whether the promised visibility translates into savings in their own operations, how security controls fit their requirements and what it costs to deploy the system in their preferred environment. Tencent’s summit announcements set out tools intended to address those concerns; the available figures do not yet establish how consistently they do so across businesses. [2], [5], [7]
Source Reference Map
Inspired by headline at: [1]
Sources by paragraph:
Source: Fuse Wire Services


