LecturaInvestigaciónRadarMarco de inversión
Iniciar sesión / Registrarse
← Volver a la lectura
NoticiasTema · 阿里AI组织转型Material3 min de lecturaEditorial de DisconfirmAI

Alibaba Executives Review AI Organizational Transformation: Why Super Individuals Don't Make a Super Organization

Jiang Fang, Chief People Officer of Alibaba Group, and Jiang Linquan, CIO of Alibaba Cloud, shared insights on AI Native organizational transformation at the Yunqi Conference. They noted that while AI improved individual coding efficiency, end-to-end delivery didn't speed up due to unchanged collaboration structures. Alibaba responded by merging R&D roles, creating innovation 'special zones,' and rebuilding system foundations, emphasizing that leaders must personally use AI to redefine workflows.

2026-09-28 16:08 GMT+8

At the Yunqi Conference on September 22, 2026, Jiang Fang, Chief People Officer of Alibaba Group, and Jiang Linquan, CIO of Alibaba Cloud, jointly discussed organizational change in the AI era. Their core argument was: "A group of super individuals does not naturally form a super organization." This insight stemmed from Alibaba's experience over the past year: although AI Coding tools significantly increased engineers' code generation rates, the end-to-end delivery speed for real requirements did not improve noticeably.

Data analysis by Jiang Linquan's team revealed the bottleneck. Developers spent only about 20% of their working time actually writing code, with the rest consumed by requirement communication, design, testing, review, and cross-role collaboration. More critically, when junior engineers used AI to rapidly produce large amounts of code, senior staff were often overwhelmed by reviewing this 'mediocre code,' causing overall efficiency to stall. This confirms the classic dilemma from The Mythical Man-Month: adding people or tools without optimizing collaboration links only increases communication costs.

Facing this gap, Alibaba began restructuring its organization. Since AI lowered the barrier to learning adjacent skills, making it possible for one person to master multiple capabilities, traditional job divisions based on scarce skills became obsolete. Consequently, Alibaba Cloud merged Product Managers, Designers, and Frontend Engineers into a PDFE role, and Backend Engineers and Architects into an ABE role, forming a new 'Half-Stack' structure. This adjustment aimed to reduce handover steps, allowing expanded individual capability boundaries to directly drive organizational division of labor.

Iniciar sesión / Registrarse
LecturaInvestigaciónRadarMarco de inversión

However, for a massive organization with over 100,000 employees, synchronized transformation is nearly impossible. Jiang Fang proposed two practical experiences: first, 'Leaders dive in first,' meaning managers must personally use AI to complete full work cycles to judge which processes should be removed; second, 'Open special zones,' allowing small teams to pilot new workflows in specific business scenarios. For instance, the Qoder team validated a new workflow by building a product in seven days with five people, while Alibaba Cloud's international team used the data-closed-loop nature of telemarketing channels as a 'self-judging' business scenario for promotion.

Beyond people and processes, underlying system adaptation was also key. Jiang Linquan found that legacy business systems were designed for humans, offering high tolerance but low speed; meanwhile, Agents' high-frequency calls and automatic orchestration could easily 'crash' old systems and pose security risks. Therefore, the team had to reverse-engineer and rebuild processes, unify business concepts, make implicit knowledge explicit, and open MCP interfaces for Agent calls. After four months of foundation reconstruction, general Agent call volumes grew 1000-fold, and digital employee products like 'RuiYiBao' (translation) and 'RuiHuBao' (outbound calling) demonstrated significant advantages in cost and efficiency.

Ultimately, the discussion returned to the human element. Jiang Fang pointed out that while AI can quickly lower skill barriers, it cannot automatically bring trust. In an era where skills are increasingly 'inflating,' the ability to define problems and values has become scarcer. Whether it's NVIDIA's emphasis on values or Anthropic's training for interviewers, these indicate that in the AI age, an organization's core competitiveness lies not just in technical execution, but in maintaining trust, embracing uncertainty, and finding its own evolutionary path.

Fuentes

  • geekpark.net↗
Tema

阿里AI组织转型

阿里巴巴集团正在经历一场以“AI Native”为核心的深层组织重构。针对“超级个体堆不出超级组织”的行业痛点,阿里首席人才官蒋芳与阿里云CIO蒋林泉在2026年云栖大会上披露了内部实践:通过合并产研岗位、设立创新“特区”及重构业务系统底座…

3 pasos1 piezasActualizado 2026-10-22 03:56

Cómo llegó hasta aquí

  1. 2026-09-16阿里巴巴CEO吴泳铭发布内部信,宣布成立“Alibaba Token Hub”(ATH),整合通义实验室、Qwen事业部等核心AI业务,直接向CEO汇报,确立AI战略的一体化运营架构。finance.biggo.com ↗
  2. 2026-09-22云栖大会主论坛,吴泳铭提出2032年全球数据中心容量超20GW的目标;蔡崇信强调AI从技术突破向价值创造转化。同日,蒋芳与蒋林泉在AI Native组织论坛分享转型实践,指出需通过岗位合并(如PDFE/ABE)和系统重构来解决协作瓶颈。geekpark.net ↗
  3. 2026-09-22阿里云发布全栈AI路线图,包括Qwen 4训练进展、真武V900 AI芯片及面向手机的Qwen Intelligence代理平台,展示底层技术与上层应用协同的组织能力。alibabacloud.com ↗

Lo que aún está pendiente

  • “Token Hub”整合后,原各事业部的独立研发节奏与统一战略之间的摩擦如何具体化解?
  • “Half-Stack”新岗位结构(PDFE/ABE)在阿里其他非云业务板块的复制效果及阻力有多大?
  • 随着Agent调用量激增,现有业务系统的安全权限体系能否持续支撑大规模自动化操作而不引发风险?
Leer el tema completo →
Continuar investigando →