On September 8, Alibaba officially launched its digital employee product QoderWake 1.0. With just one sentence describing a job requirement, QoderWake generates a role document. Once confirmed, it "hires" a digital employee that understands job responsibilities, collaborative relationships, and work permissions.
Over the past three months, nearly 100,000 digital employees have been "on the job" in real work scenarios, executing approximately 2 million effective tasks. Half of these tasks were triggered automatically by time or events, starting work before humans even spoke.
Hire a Digital Colleague in Minutes
QoderWake comes with 10 pre-trained positions: frontend, backend, product manager, content operations, data analyst, project administrator, UI designer, DevOps engineer, and the Q&A specialist "QZai"—covering R&D, design, project, data, content, and technical support. Each position is fine-tuned with best practices from industry-leading teams, ready to handle real tasks without additional training.
Users can also customize digital employees: define core responsibilities, working style, workflows, and guardrails, then attach business documents to "train" a dedicated digital employee. Version 1.0 further simplifies the process—describe the role in one sentence, the system generates a draft for review, and once confirmed, it can be saved as a template for future use.
Collaborative by Default: @ Them in DingTalk or Feishu
Each digital employee in QoderWake is called a Waker. Version 1.0 upgrades Wakers from "drop a task and wait for a reply" to staying in group chats and participating in discussions.
Wakers continuously process new messages, determining whether they're follow-ups, corrections, or new tasks—then decide whether to handle them, pass them to another Waker, or bring them back to a human. Attachments and confirmed decisions from DingTalk and Feishu group chats become continuously updated knowledge for the digital employee.
Triggers go beyond @mentions: scheduled rules, keyword matches, and API events can wake Wakers. Digital employees "work alongside people during the day and keep working through the night," 24/7.
Self-Evolution: 276,000 Memories, the More You Use It, the Better It Knows Your Organization
After each execution, experience is structured into five dimensions: memory, skills, strategies, validation rules, and workflows. As of September 2026, the platform has accumulated 276,000 memories and 123,000 skills, of which 22,800 came from self-evolution.
A built-in "anti-corruption governance" mechanism deduplicates, resolves conflicts, and removes outdated historical experiences to prevent the system from learning bad patterns. Inside Alibaba, the Q&A digital employee "QZai," built on QoderWake, has served 100,000 employees with a 72% accuracy rate.
Multiple Wakers can form "groups" to collaborate: a product manager Waker drafts requirements, a developer Waker writes a spec, humans confirm, then development and testing are completed, with test cases generated and executed by a QA Waker.
Security Guardrails: High-Risk Operations Blocked by Default
Before going on the job, digital employees have clear boundaries: sensitive data stays local, high-risk operations are blocked by default, and predefined rules trigger authorization requests to users. Each Waker comes with a permission center, with four dimensions—tools, files, built-in capabilities, and model outputs—controllable independently.
On the commercial front, according to 36Kr, Alibaba's Qoder suite ARR has exceeded $60 million, placing it among the top tier of domestic Agent products. The pricing model is shifting from traditional "per-seat" to "task or role-based output"—valuing work completed or business results delivered.
QoderWake's most significant upgrade turns AI from a "tool to be invoked" into a "colleague with a desk." The combination of four capabilities—joining group chats, autonomous triggering, experience accumulation, and team formation—points to a more profound shift: AI is moving from "helping you work" to "standing in for you while you're away." When digital employees can monitor CI pipelines around the clock, answer customer questions, write code, and self-correct, the nature of human-AI collaboration within organizations is being redefined. The nearly 100,000 digital employees already on the job are validating not just technical feasibility, but the real-world boundaries of a new organizational paradigm: human-machine synergy.