Tidong Wei
R\&D Director at Sangfor, Head of Costrict
Head of Costrict: Graduated in 2009 with a degree in Software Engineering from Wuhan University. Since joining Sangfor, he has served as a developer, architect, operations lead, and R\&D director, gaining extensive expertise in security, cloud computing, big data, and AI deep learning. Current role: Internally, he is responsible for driving efficiency across the company’s entire R\&D process through AI; externally, he leads the Costrict product—an open-source AI programming tool designed for enterprise-grade development.
Topic
From Vibe to Costrict: How AI Programming is Transforming Enterprise-Level Development
While many focus on “ambient programming,” using AI to achieve the smooth experience of generating code in minutes for personal projects, enterprise-grade development faces far tougher challenges: AI can generate code quickly, but meeting the deployment standards of “intranet data security + high quality” is extremely difficult. In serious development scenarios, every line of code carries clear objectives and responsibilities—transaction security in financial systems, order accuracy on e-commerce platforms, compliance in medical software—leaving no room for the gamble of “deploy first, fix later.” This talk will focus on the core idea that “quality is the foundation of efficiency in serious scenarios,” unpacking the traditional development dilemma of speed versus stability, and demonstrating with real cases and data the fatal costs of deferring issues. It will show how enterprise-grade development can embed quality throughout the AI development lifecycle—from “setting rules before coding,” to “self-validation during coding,” and “double-checking after coding.” By making quality the starting point, AI can drive development that not only addresses intranet security and controllable quality but also truly improves efficiency by reducing rework, providing a new solution for serious enterprise development.
