Ant Group releases Lingguang 2.0, positioned as a personal AI assistant that can continuously manage tasks

The flight and hotel were booked long ago, yet I still can't resist checking the prices every few days. If they've dropped, I calculate the cancellation and change fees to see whether I should rebook; if not, I check again in a couple of days. It doesn't take much time, but it's always on my mind.

I signed up for a membership to binge a show, stopped opening the app once I finished it, but the auto-renewal never stopped. It wasn't until I went through my bills one day that I realized several more months of charges had gone through.

On October 27, Ant Group released Lingguang 2.0, positioned as a "personal AI assistant that can manage things," making it one of the first Personal Agent products launched by a major Chinese tech company.

The change it aims to bring is straightforward: no longer passively responding to commands and completing one-off tasks, but proactively identifying needs and continuously helping you manage affairs.

After Muse, global competition in Personal Agents began to accelerate, and major Chinese companies are also laying out plans intensively. Lingguang is among the first to deliver a product answer, starting with everyday matters like watching ticket prices and sorting out bills, letting AI begin to continuously manage things for users.

From "one-time help" to "long-term management," Lingguang moves to its next stage

When Lingguang 1.0 launched last year, what left the deepest impression was how it delivered answers—a user makes a request, and within 30 seconds Lingguang hand-builds an interactive mini-app with 3D and charts, letting many ideas quickly take shape.

But after delivering an answer or an app, the service often came to an end. When needs changed, information updated, or plans needed adjustment, users had to come back and explain again.

Lingguang 2.0 turns the service from a "sprint of doing errands" into a "long run of managing affairs." When users bring matters that need ongoing attention into "Manage for Me" during a conversation, Lingguang can keep following up.

Take Leiphone.com as an example: editors need to continuously follow product launches, technology updates, and industry discussions from overseas AI companies. But news events are extremely numerous, information is complex, and trends change very quickly, with the same event generating a large number of retellings. Watching it yourself takes a lot of time, but not watching is not an option.

Matters that require ongoing attention like this can be handed directly to Lingguang, letting it filter out events worth editors' further verification and tracking. In "Manage for Me," you can see the corresponding overseas AI hotspot monitoring task, and click in to view the progress of information collected each day.

Matters in daily life can also be left to Lingguang.

Recently, Shenzhen advocated paid leave for companies during the APEC meeting, and this editor was also itching to take the kid out for a trip, so I asked Lingguang to check whether there were still cheap flights from Shenzhen to Beijing and from Shenzhen to Harbin.

At the time, Lingguang told me that according to the records it found, ticket prices had already started rising on Saturday, and suggested I grab a suitable one right away. But I didn't believe it and asked it to keep watching.

Two days later, Lingguang reported that low-price tickets to Beijing continued to jump up in tiers, while Harbin ticket prices were relatively stable, and suggested buying as soon as possible.

Beyond these long-cycle tasks, Lingguang can also help analyze my Alipay bills. Previously, Huabei on Alipay had a pile of bills every month, and I was always too lazy to check them. After Lingguang came out, I also asked it to carefully break down the bill details for me.

Now, Lingguang's management capabilities have entered my daily life. For example, if I want to buy a Siemens dishwasher with a budget of around 6,000 yuan, it can continuously compare prices on JD, Taobao, Pinduoduo, and Douyin; if the kid wants to start learning English, it can help make a course plan.

From goals users state, to needs gradually formed during conversations, to issues not yet noticed in authorized information, the affairs Lingguang tries to take on are beginning to multiply.

Over the past year, Agents such as Claude Code and Codex have shown people AI's ability to execute complex tasks. Users state a goal, and the system autonomously plans steps, calls tools, and executes very complex tasks, with completing a specific task as the objective.

But in users' actual lives, a large number of needs have no clear end time, and external conditions keep changing.

As a Personal Agent, what Lingguang is trying to take on is exactly this segment: AI can remember the ins and outs of a matter over the long term, continuously follow subsequent changes, and proactively take action when needed, so users can watch fewer things and worry less.

The upgrade of Lingguang 2.0 comes during a window when global AI companies are re-searching for the product form of personal assistants.

In September, Meta launched Muse, organizing long-term memory, background work, and cross-app execution into a single product, and Meta subsequently introduced connections to services such as shopping and travel, expanding the range of needs the assistant can take on.

Soon after, Chinese companies also accelerated: ByteDance advanced a personal agent project codenamed Spell, planned to be combined with Doubao; Tencent launched a cloud-based personal assistant, LightVela.

Overseas users' information and services are scattered across email, calendars, independent websites, and subscription accounts, so Muse emphasizes a delegation relationship of "autonomous execution." For example, if a user wants to lower a service bill, they can hand the goal to Muse, which uses a browser to find options and advance the relevant operations, contacting the user when a step requires approval.

Foreign users' information and services are scattered across various apps and mini-programs; single operations are already convenient, but cross-platform information and long-term matters still require users to keep track themselves.

So what Lingguang chooses is to take over matters scattered across different platforms that need to be remembered long-term and checked repeatedly, so that after users hand them over, they can worry less.

What supports this product at its core is the coordination of cross-session memory, long-horizon task scheduling, and tool connection capabilities in Lingguang 2.0.

Take the example of watching for low-priced flights: cross-session memory lets Lingguang remember travel dates, budget, and time-of-day requirements; long-horizon task scheduling lets it continue checking at suitable later times, while tool connections help it obtain new flight and price information.

Memory is responsible for remembering clearly, scheduling for following through, and tools for getting new information. The system then uses this information to judge which changes are worth reminding the user about. An independent cloud runtime environment lets this work continue after the user exits the chat.

Users do not need to introduce the situation from scratch every time, nor repeatedly come back to ask for an update.

Lingguang 2.0's advantage also lies in its technical ability to continuously manage matters, connecting to China's distinctive digital services ecosystem.

It adopts two complementary approaches:

Some common services are built in directly, including Alipay bills, memberships, and travel, as well as DingTalk, Feishu, email, Mijia, and others, reducing users' configuration costs; more scattered needs continue to be extended through independent computers, browsers, and open tools such as MCP and CLI.

Built-in services let the product start from everyday needs, while open tools leave room for long-tail tasks.

Lingguang can therefore start from services familiar to Ant and, through connections to office platforms, email, and smart devices, extend beyond Alipay. It tries to organize tools around one matter of the user's, rather than confining work inside a single app.

From "managing matters" to "entrusting": where Ant's advantage lies

Ant's long-accumulated capabilities in payments and life services provide the foundation for better connecting services and users.

At present, Ant already has a large number of merchant collaboration relationships and business experience; Lingguang can advance these transformations along the existing network without needing to find service providers anew and establish cooperation for each need. This is the practical advantage that the digital services ecosystem can translate into.

After AI helps users find suitable products, payment needs to be completed; when merchants provide services to AI applications, they need to receive payment; and when developers open up tools and capabilities, they also need metering, reconciliation, and settlement. Alipay's AI payment layout is covering agent payments, AI application payment collection, and metering and settlement of tool calls, providing support for these needs.

For a personal agent to enter the daily lives of Chinese people, it needs to connect cutting-edge technology to local account systems, applications, and service networks, and establish service relationships that users are willing to entrust over the long term.

Being able to link these elements together is precisely the differentiated advantage Ant has formed through years of accumulation.

Entering the Personal AI era, the way people connect with services has new possibilities.

Business competition will also extend from models and individual applications to the coordination of technical capabilities, service ecosystems, and user trust.

So Lingguang is an important entry point for observing changes in the AI Agent industry. The "continuous management of matters" it explores has the opportunity to extend the connections Ant built around transactions in the past into services organized around users' long-term needs.

The commercial imagination of Personal AI comes precisely from this more sustained user relationship.

For Ant, this is an opportunity to further activate years of accumulated service ecosystem; for ordinary people, the value is more plain: those small matters that have always been on their minds are remembered and followed through by someone, so they can worry a little less.