Ant’s LingGuang 2.0: An Assistant That Works While You’re Away

Human Verified 0

Share to: , and

A person walking through a bright city plaza glancing at a phone with a blank calendar screen

Ant Group has upgraded its LingGuang assistant into a proactive personal agent that keeps working after the user closes the chat, spotting needs, monitoring changes and following up over time. LingGuang 2.0, rolling out gradually in China this week, runs on independent cloud instances with cross-session memory, and connects to Alipay, DingTalk and other everyday services with the user’s explicit authorisation.

From answering to managing

The core change is the starting point of the work. Instead of waiting for a command, LingGuang can, with explicit user authorisation, draw on connected information and services to spot potential needs, initiate analysis and follow-ups, and alert the user when attention is required, moving from one-off execution to sustained task management.

That shift is powered by upgrades to cross-session memory, long-horizon task scheduling and tool-calling. Because the agent runs on independent cloud instances, tasks continue beyond a single conversation session, dynamically following up as new information and status changes arise, rather than restarting from scratch each time.

A cosy modern living room at dusk with warm lamps and a phone on a coffee table

Supported uses show the ambition. With authorisation to access Alipay, LingGuang can review spending records, identify potentially forgotten auto-renewing subscriptions, analyse spending patterns and offer savings recommendations. After flight or hotel bookings, it keeps monitoring prices and notifies the user when a lower price appears.

Connected, with permission controls

LingGuang connects to Alipay, DingTalk, Feishu, email and Xiaomi’s Mijia smart-home platform, and supports external tools through MCP and command-line interfaces. It can extract tasks from school notices in DingTalk groups, extending its reach from payments and travel into family and household routines.

Security rests on Ant’s APASS account system, with authorisation keys held separately from the agent, so users can view, adjust or revoke permissions at any time. Agents are meant to operate strictly within authorised boundaries, and that permission model is the practical answer to the obvious question about an agent with access to payments and private messages.

LingGuang 1.0 launched in November 2025 and reached 2 million downloads in six days. TechNode notes that version 2.0 is the first personal-agent product to market among China’s major technology companies, which makes its design choices a reference point for rivals building similar agents.

LingGuang 2.0 is rolling out in China on iOS, Android and the LingGuang website, after limited testing earlier in October. Ant has not stated availability outside China, and the examples in its announcement all concern Chinese services, from Alipay billing to DingTalk school notices and Mijia smart-home devices.

Why It Matters

The personal agent is the next step beyond the chatbot, and LingGuang 2.0 shows what that step actually requires: memory across sessions, scheduling across days, and permission to act in connected services without asking again at every step. An assistant that only answers is a tool; one that follows up is closer to a delegate.

That delegation is also the risk. An agent that can see payment records, private communications and home devices is useful precisely because it sees so much, and commentators have raised the unresolved privacy question around data held under China’s legal framework, which Ant’s permission controls cannot override. Availability outside China has not been stated.

The comparison that matters for readers elsewhere is the assistant acting on personal data, as with ChatGPT’s mail and calendar plugins, covered in our ChatGPT iCloud plugins report. Every platform faces the same bargain: more access means more useful agency, and more trust to lose if the boundaries fail.

Background

Safe execution is the other half of the agent story, with vendors building guardrails for agents that act in the world, as covered in our Akamai agent execution report, while consumer agent deployments are spreading, as covered in our Concentrix voice translation report. LingGuang 2.0 is rolling out in China, so its behaviour will be watched closely.

Source: Ant Group announcement TechNode report

Share to: , and

Faheem R. Khattak
Founder & Senior Editor

Senior automation engineer and AI marketing journo covering LLMs, algorithms, analytics, core updates, Bing search, and technical SEO.

✓ 10+ Years Covering Google Search & AI

0 Comments

Leave a Reply

Optional Take Tag:

🛡️ Spam protection: Honeypot + Time-Trap Active

This site uses Akismet to reduce spam. Learn how your comment data is processed.