The figures reported by Ant Group
Ant Group has released the first data on the new AI assistant Lingguang 2.0, a version distributed since October 27. According to IT Home, the source of the information, the platform handled more than 1.16 million activities for users during the first week, executing over 3.6 million tasks. During the same period, the service's user base exceeded 20 million.
These are figures communicated by the company itself and not independent measurements: the material does not clarify how managed activities or executed tasks are counted, nor whether the latter are included in the activities or constitute a separate category.
What changes in version 2.0
The main novelty claimed by Ant Group concerns the starting point of the AI's intervention, which moves further upstream. In the previous model described by the company, the user had to identify a problem and formulate a request before the assistant intervened; with Lingguang 2.0, after obtaining authorization, the system can combine information and services already connected to recognize potential needs, launch analyses, monitoring and follow-ups, and alert the user at the moments when their attention is needed.
The stated consequence is a change of role: the assistant does not limit itself to waiting for instructions, but becomes a tool that follows activities continuously. The material does not indicate which services are actually connected or in which scenarios the system has launched initiatives autonomously.
Security and authorizations via APASS
Lingguang 2.0 also integrates APASS, described by Ant Group as a unified account and security infrastructure for the AI era, designed to work across agents, devices, applications and different service scenarios. APASS embeds authentication, intent security and privacy protection into the agent execution flow, with mechanisms such as separate custody of authorization keys, so that the agent can operate without coming into direct contact with the keys.
According to the company, users can at any time consult, modify or revoke authorizations, keeping the agent within the granted limits. The practical effectiveness of these controls and the ways in which data collected for proactive identification of needs are handled remain to be verified.
