Urgent.News

650+ sources. One page. See who else covered it.

Editions

AI

I gave Tencent’s WeChat AI agent control for 24 hours: where it excelled – and stumbled

Tencent Holdings’ WeChat is already the undisputed super-app for over a billion people in China, handling everything from splitting dinner bills to hailing taxis. Now, Tencent is embedding an artificial intelligence agent named Xiaowei into the platform, adding a hands-free, automated assistant to its vast digital ecosystem. In its latest earnings results announced on Wednesday, Tencent…

I gave Tencent’s WeChat AI agent control for 24 hours: where it excelled – and stumbled

Tencent's WeChat AI agent, Xiaowei, was tested by a reporter for nearly 24 hours to assess its capabilities. Unlike standard chatbots, Xiaowei is designed to perform actions, such as making reservations and placing orders. In my experience, Xiaowei successfully helped with tasks like sending messages, initiating audio calls, setting reminders, and retrieving information from WeChat.

It also independently navigated mini-program menus to place food orders, requiring only the final confirmation. However, Xiaowei fell short in certain areas, including slow performance, misheard names, and difficulties in placing voice calls. Additionally, the AI-assisted writing function for WeChat Moments drew criticism due to concerns about diluting genuine human connection.

Developed using WeChat's in-house AI model, WeLM, Xiaowei remains constrained by common AI agent issues like sluggish execution and errors. Despite these limitations, Tencent's advantage lies in its ecosystem where social networking, payments, and daily services are deeply integrated, providing a strong foundation for Xiaowei's potential future growth.

Written by urgent.news from South China Morning Post's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Also reported by 1 other outlet

Read the original at scmp.com →

More in AI

I Built an Investigation-First Debugger for AI Agents — TraceMotive v0.3

I Built an Investigation-First Debugger for AI Agents — TraceMotive v0.3 I just released TraceMotive v0.3.0 . TraceMotive is an open-source, local-first debugging tool for AI agent executions.

  • TraceMotive is an open-source debugger for AI agent investigations
  • Focuses on deterministic findings like tool changes and execution subtrees
  • Prioritizes privacy with loopback-only data persistence and no analytics

More from Sunday 16 August →