How AI Helps Teachers Improve Efficiency
QwenWork Case
After finishing classes for the day, Wang Yize sends Qwen Work a batch of audio recording links from that day's lectures.
"For example, if I taught five classes today, I just need to copy and paste five links, and it will read them automatically."
Behind each link lies an actual session of his "Western Economics" course. The transcripts contain knowledge points covered by the teacher, real-time examples given during class, and how students responded. Previously, this data was scattered across individual sessions; now, Wang Yize has started using Qwen Work and DingTalk AI Education Edition to connect all these pieces together.
First, save one class session
Wang Yize is a teacher specializing in college entrance exam preparation. He teaches economics-related courses year-round, instructing as many as 2,500 to 3,000 students annually.
The more mature a course becomes, the more stable its knowledge framework should be—making a teacher’s workload seemingly lighter. But Wang feels precisely the opposite. "When you’ve refined a lesson to perfection, your entire career becomes mere copy-and-paste."
The same content must be re-explained repeatedly to new batches of students every term. Yet once class ends, spontaneous examples given on the spot or areas where students struggled are difficult to preserve completely.
If students don’t understand something, they either have to ask again or go back and manually scrub through video recordings. Meanwhile, teachers still need to prepare lesson plans, write exercises, grade assignments, and figure out exactly where students got stuck.
A month ago, Wang began using the AI Education Edition, and this “copy-and-paste” cycle started changing. After each class, besides converting audio into text, the AI Education Edition automatically generates summaries, outlines, and lecture notes. When students miss a concept, they no longer need to replay the entire recording searching for answers—they can simply refer back to structured notes containing key concepts, difficult points, and example analyses.
Wang had experimented with large language models to generate economics practice questions long before, but when asking the model to create questions targeting specific chapters or知识点 (knowledge points) in Western Economics, it often produced off-topic or unbalanced difficulty levels. To build a usable question bank, he had to repeatedly feed in old questions, textbooks, and classroom case studies.
Now, the AI Education Edition assists in generating questions based directly on what was actually taught in class—the real content and emphasis—and allows Wang to perform secondary editing afterward.
"Because large models don't know me. But after the AI Education Edition listens to my class, it gains sufficient context and information to understand the focus, syllabus, and content of the lesson. So the questions it generates are more accurate, and I feel much more confident about them."
Student performance data is also clearly recorded. Within the AI Education Edition platform, teachers can see how long students hesitated on each question by analyzing response times, identifying which knowledge points remain unclear—providing valuable insights for personalized teaching later.
According to Wang’s experience, tasks that used to take an entire afternoon—organizing lesson plans, writing exercises, summarizing errors—now take roughly 30 minutes. Class-wide common mistakes that previously took two days to identify become visible immediately after exercise submission.
Voice recordings, examples, and answer processes preserved from each class provide solid references for future question creation and explanations.
Then, connect multiple class sessions together
Wang might teach four or five classes a day across several different groups. Now that every class is recorded, he begins paying attention to long-term trends: What weak knowledge points keep appearing over a month? How do learning patterns shift across different classes? Which classroom content deserves to be turned into new teaching materials or question banks?
These questions go beyond the scope of any single class.
In Wang’s workflow, the AI Education Edition handles recording and analyzing individual sessions, while Qwen Work takes over the multiple transcripts he submits, enabling longer-term educational research perspectives. After a day’s teaching, he sends several links at once—allowing previously isolated knowledge points, examples, and teaching expressions from various classes to be processed collectively.
These records continue accumulating. After one month, one semester, or even longer, Wang hopes to analyze how teaching priorities evolve and which concepts consistently challenge students. Such long-term conclusions will require time to validate.
Qwen Work also takes care of some repetitive daily tasks.
Wang manages multiple classes, each with their own DingTalk group. After assigning homework, he used to remind students or parents individually in each group and then follow up on submissions. Now, he has set this process as a scheduled task in Qwen Work. Every weekday at 7 p.m., Qwen Work automatically reminds students or parents, and Wang follows up afterward.
Classroom records are now being used to organize teaching materials and exam papers. Single-class question generation is handled by the AI Education Edition, but when preparing full teaching materials or exams, Wang feeds existing textbooks or reference templates into Qwen Work.
This workflow has been formalized into two Skills: "Textbook Setup" and "Exam Paper Setup."
Making AI increasingly familiar with his teaching style
This working mode has also reshaped Wang’s understanding of “AI-assisted teaching research.”
He divides his journey with AI into three stages. Initially, it was “AI can’t do it”—models could generate questions but frequently went off-syllabus or made errors. Then came “AI doesn’t know me”—he had to repeatedly input textbooks, past questions, and course background. Now, what he wants is “AI that is both useful and knows me.”
This “knowing” comes from real classroom sessions, from the content teachers deliver and the questions students attempt. The AI Education Edition preserves this information, which Wang then hands over to Qwen Work as contextual foundation for further organization and pedagogical research.
Wang believes that using AI does not merely improve efficiency in generating teaching materials, exams, or analyzing classes—it significantly reduces the effort required to move information between workflows.
Previously, he had to find recording devices, transcribe software, and manually copy materials into models one by one. Each time he switched tools, he had to re-explain the course content, syllabus, and难点 (difficult points). No matter how powerful the model, without proper context, it always had to start fresh with him.
Now, the AI Education Edition first captures authentic classroom data, and Wang forwards the transcription links to Qwen Work. By learning this course content, AI becomes capable of participating in long-term teaching research.
"I think the strength of Qwen Work—or what makes it truly important—is its ecosystem, interaction, and interconnectivity," says Wang.
Wang’s approach reflects an emerging AI trend. For agents to effectively take on tasks, they must have access to rich contextual data. In the future, more hardware devices may serve as sources of context for AI agents, capturing and recording information from classrooms, meetings, and everyday work.
We dedicated to serving clients with professional DingTalk solutions. If you'd like to learn more about DingTalk platform applications, feel free to contact our online customer service or email at
Using DingTalk: Before & After
Before
- × Team Chaos: Team members are all busy with their own tasks, standards are inconsistent, and the more communication there is, the more chaotic things become, leading to decreased motivation.
- × Info Silos: Important information is scattered across WhatsApp/group chats, emails, Excel spreadsheets, and numerous apps, often resulting in lost, missed, or misdirected messages.
- × Manual Workflow: Tasks are still handled manually: approvals, scheduling, repair requests, store visits, and reports are all slow, hindering frontline responsiveness.
- × Admin Burden: Clocking in, leave requests, overtime, and payroll are handled in different systems or calculated using spreadsheets, leading to time-consuming statistics and errors.
After
- ✓ Unified Platform: By using a unified platform to bring people and tasks together, communication flows smoothly, collaboration improves, and turnover rates are more easily reduced.
- ✓ Official Channel: Information has an "official channel": whoever is entitled to see it can see it, it can be tracked and reviewed, and there's no fear of messages being skipped.
- ✓ Digital Agility: Processes run online: approvals are faster, tasks are clearer, and store/on-site feedback is more timely, directly improving overall efficiency.
- ✓ Automated HR: Clocking in, leave requests, and overtime are automatically summarized, and attendance reports can be exported with one click for easy payroll calculation.
Operate smarter, spend less
Streamline ops, reduce costs, and keep HQ and frontline in sync—all in one platform.
9.5x
Operational efficiency
72%
Cost savings
35%
Faster team syncs
Want to a Free Trial? Please book our Demo meeting with our AI specilist as below link:
https://www.dingtalk-global.com/contact

English
اللغة العربية
Bahasa Indonesia
日本語
Bahasa Melayu
ภาษาไทย
Tiếng Việt
简体中文 