Ant Group Releases Lingguang 2.0, AI Assistant Shifts to Proactive Task Management

After booking a flight, AI watches the price around the clock and alerts you promptly when it drops; when reviewing your everyday bills, AI identifies subscriptions that may be idle renewals and helps you save money; when parents are too busy to check the class group chat, AI can also sort through the teacher's notices and extract to-do items... These everyday chores that users previously had to discover, remember, and arrange for themselves are now beginning to be taken over proactively by AI.

On October 27, Ant Group released Lingguang 2.0, driving the AI assistant to evolve from passively responding to instructions and completing single tasks toward proactively discovering needs and continuously managing affairs.

Compared with traditional conversational AI and task-oriented Agents, one major change in Lingguang 2.0 is that it moves the starting point of AI involvement in tasks even further forward. In the past, users first had to discover a problem and raise a request, and then AI would execute it; now, after obtaining user authorization, Lingguang can combine the information and services it has access to, proactively identify potential needs, initiate analysis, monitoring, and follow-up, and promptly remind users at the points that require their attention. AI is no longer just a tool waiting for instructions, but is beginning to become an assistant capable of continuously managing affairs.

Behind this change is Lingguang 2.0's comprehensive upgrade of cross-session memory, long-horizon task scheduling, and tool connectivity. Relying on an independent cloud host, Lingguang can break through the limits of a single conversation, continue running tasks after the user exits the conversation, and dynamically follow up based on new information and changes in status, shifting AI from one-off task execution to long-term, continuous affairs management.

Lingguang's ability to manage affairs is now entering specific real-life scenarios.

In the consumption field, after the user authorizes access to Alipay, Lingguang can sort through consumption bill records, discover automatic renewal subscriptions that may have been forgotten, analyze spending habits, and provide money-saving suggestions; in the travel field, completing a flight or hotel booking does not mean the AI service ends—Lingguang can continue to monitor subsequent price changes, discover price-drop opportunities, and promptly notify the user; in family scenarios, after authorization, Lingguang can follow teachers' notices in DingTalk class groups, organize important information, extract to-do items, and complete follow-up work such as document formatting.

In addition, for matters that require long-term memory and continuous follow-up—from pet care and children's vaccinations to family anniversaries and vehicle inspections—Lingguang can also proactively identify key nodes within the scope of authorization, arrange follow-up, and reduce the burden on users of repeatedly checking, reminding, and following up.

From completing tasks once to proactively understanding and continuously advancing them, this means the AI Agent is shifting from a "task executor" to an "affairs manager." To achieve this capability, model understanding and reasoning alone are not enough; it also requires cross-application information acquisition, tool invocation, and infrastructure support for stable long-term operation.

According to the introduction, Lingguang 2.0 not only connects the data and services within the Alipay ecosystem, but also supports calling tools such as MCP and CLI in open environments, expanding cross-platform collaboration capabilities. At present, Lingguang has built in service capabilities including Alipay bills, memberships, and travel, as well as DingTalk, Feishu, email, and Mijia, and its management scope is extending from personal information and online affairs to cross-platform workflows and smart home device control.

As agents move toward "affairs management," their permission boundaries, data security, and execution credibility have also become new challenges. Lingguang 2.0 has integrated Ant APASS as the unified account and security management infrastructure for the AI era across agents, devices, applications, and service scenarios. APASS incorporates identity authentication, intent security, and privacy protection into the Agent execution process, and through mechanisms such as separate custody of authorization keys and the Agent, allows the agent to execute tasks without directly touching the authorization keys. Users can view, adjust, or revoke authorization at any time, so that the agent acts within the scope of authorization.

It is understood that the Lingguang App is a full-modal general-purpose AI assistant launched by Ant Group, officially launched on November 18, 2025, with downloads exceeding 2 million within six days of launch. Previously, Lingguang, with features such as "generate a flash app in one sentence" and "generate your world," extended natural language interaction to app generation and multimodal content creation, leading a new interaction model for AI assistants.

The real breakthrough of a personal Agent is that AI begins to have a time dimension. Chatbots mainly solve immediate problems, while personal Agents need to handle matters spanning hours, days, or even longer cycles. It must remember context, perceive changes, advance tasks, and re-request human judgment when necessary.

This upgrade of Lingguang also reflects the next stage of competition for AI assistants: compared with how many questions it can answer and how many instructions it can complete, whether AI can understand users' long-term needs, proactively intervene at the right moment, and continuously manage things well is becoming a new dimension for measuring whether agents can enter real-life scenarios.