On June 15, WeChat Pay officially released the AI Access Toolbox 2.0. This version, building upon the initial product launched in April (including Skill packages, AI-friendly documentation, and AI-friendly APIs), has completed five major technical upgrades based on practical feedback from merchants and developers, focusing on three core directions: security, efficiency, and native smoothness. The aim is to completely resolve the concerns of over 70% of merchants about fund security when using AI-assisted programming, further reducing the development threshold and resource consumption.
In this iteration, the toolbox has achieved native language adaptation globally, fully supporting nine languages including Chinese, English, Japanese, and Korean. It also performs native semantic calibration for WeChat Pay's specific business concepts, eliminating translation errors caused by direct machine translation.
At the same time, the comprehensive knowledge base has been deeply expanded, covering all WeChat Pay products. In terms of capabilities, the version 2.0 introduces a dual AI expert matrix: "Technical Experts" and "Financial-grade R&D Experts." The former handles end-to-end Q&A and troubleshooting, while the latter reviews code according to financial-grade security standards to avoid potential vulnerabilities.
In addition, the new version introduces for the first time the CLI dynamic troubleshooting function, allowing developers to check order status in real-time through natural language without leaving the coding environment. The document charts have also been uniformly converted to Mermaid format, combined with an automatic update mechanism at startup, significantly improving the efficiency of large model reading and reducing Token consumption by 50% compared to original HTML documents.
From the initial trial of the toolbox to the comprehensive evolution of version 2.0, WeChat Pay is upgrading the traditional API integration model into AI-native interaction through the reconstruction of the toolchain. This not only greatly improves the R&D efficiency and fund security of millions of merchants but also marks that the application of large models in China is accelerating from "generalized conversation" to "vertical industry workflow" deep penetration.
