Português
EnglishWhether enterprise-level Agents can truly be implemented depends on the integration and application of data.
QwenWork Event
At the 2026 Yunqi Conference held on September 22, Qwen Office officially launched a new product designed for enterprises—Enterprise Context.
This product transforms an enterprise's internal knowledge base and business data into context information that Agents can access in real time. Combined with Qwen Office's powerful Agent hosting capabilities, it helps companies build digital employees that truly "understand the business and can collaborate."
The following is a summary of the on-site presentation:
I'm very honored to share with you today. Qwen Office is an AI Agent solution focused on enterprise scenarios. Perhaps many of you have attended numerous AI product launch events, but I doubt you've ever seen someone take the stage holding a cute little octopus-shaped microphone—this is our brand mascot. We hope that while promoting efficient work, we can also bring a bit of lightheartedness to the workplace. We believe that AI not only boosts efficiency but should also give people more time to enjoy life and stay happy.
Context is All You Need
Many enterprises face real challenges when pursuing AI transformation. What exactly are these difficulties? We've identified three core issues.
First, today's AI is already quite intelligent and can handle most common conversations or tasks. However, it still cannot answer questions like, "What was our company's gross profit margin in East China last quarter?" because it simply doesn't know which company "our company" refers to. If an Agent doesn't understand the specific details of how a company operates, it will struggle to integrate into actual workflows.
Second, although individuals see significant efficiency gains after using AI—some employees can even finish a day's work in half a day—these results are hard to replicate across the entire organization. Without an effective mechanism for capturing and sharing best practices, excellent experiences cannot be quickly scaled.
Finally, while companies want to adopt AI, they often hesitate due to concerns about data leaks. How to ensure data security while reducing costs is currently the most critical issue.
The Foundation of Successful Enterprise-Level Agents: Effective Data Utilization
Only by properly integrating and managing data can Agents become deeply embedded in core business operations. We propose three key elements: connection, understanding, and sharing.
"Connection" means linking data sources from various systems within the enterprise; "understanding" involves intelligently processing and condensing this integrated data; and "sharing" ensures that individual best practices quickly become assets available to the entire organization.
Qwen Office: Business-Savvy, Collaborative, Trusted
We aim for Qwen Office to be the preferred platform for enterprises adopting AI Agents, characterized by three qualities: business-savvy, collaborative, and trustworthy.
"Business-savvy" means that Agents can grasp the unique terminology and rules of each company—for example, clearly knowing which provinces comprise the "East China region" or how "gross profit margin" is calculated. These insights aren't guesses—they're built through direct access to a company's real-world data.
"Collaborative" implies that a successful sales pitch by a top performer can instantly become a shared script template for the team, enabling immediate replication of best practices.
"Trusted" focuses on data security. Having previously founded a cybersecurity company, I fully understand the stringent security requirements of businesses. That's why Qwen Office has built a comprehensive security framework from the ground up, allowing companies to use it with confidence.
Customer Case: Guming Tea
Guming operates over 10,000 stores, and whenever a new product launches, standardized preparation instructions must be communicated quickly. Previously, they relied on training videos, but finding the right clip was difficult, and if a staff member forgot a step, they had to ask colleagues for help.
Now, with Qwen Office's Enterprise Context feature, all internal knowledge—such as ice machine filter replacement schedules, the tea-to-milk ratio for guava milk tea, and troubleshooting procedures for brewing machines—is systematically organized within the system.
To external AIs, this information might seem meaningless, but to the company itself, it's crucial. Only when Agents truly understand these details can they provide real value.
New Product Launch: Enterprise Context
Today, we officially introduce the "Enterprise Context" feature, seamlessly connecting corporate knowledge with Qwen Office.
Whether it's years' worth of documents, meeting notes, group chats, or even tacit knowledge stored in employees' minds, everything can be linked and integrated into the Enterprise Context.
Whenever a task is executed, Qwen Office automatically pulls relevant information from this context and completes the job according to the company's established standards.
Key Challenge: Vast Data Volume vs. Context Length Limits
On average, medium to large enterprises possess around 150 PB of internal data, yet even the most advanced models today support only millions of tokens of context. Clearly, you can't just dump all that data in.
Piling in too much low-density information not only risks exceeding capacity limits but also wastes precious context space.
Solution: Connection + Intelligent Compression
We must effectively link and condense enterprise data to increase its information density.
Leveraging large model capabilities, we can map out the entities within a company and their relationships—similar to how Palantir builds ontologies—but at lower cost and faster speed.
Data is progressively compressed by layer, loading only what's needed when required, much like a file system, to avoid resource waste.
We also employ graph database technology to connect projects, people, and documents, giving Agents a deeper understanding of the company's operational landscape.
Powerful Model Support: Full-Stack Technical Backbone from Qwen
Qwen Office is backed by Alibaba's complete AI technology ecosystem, particularly the Qwen series of models.
When dealing with massive multimedia files (like images and videos), Qwen3.8-Omni-Flash efficiently parses and compresses unstructured content, incorporating it into the Enterprise Context.
And based on the cost-effective Qwen3.8-Flash language model, we continue optimizing data update efficiency to keep pace with daily changes in corporate environments.
Open Ecosystem: Full Support for DingTalk, Feishu, and WeChat Work
Qwen Office has achieved seamless integration with popular collaboration platforms like DingTalk, Feishu, and WeChat Work.
No matter which tool a company prefers, we strive to deliver a consistent user experience. Our philosophy is openness and consistency.
Only by remaining fully open can we serve diverse industries in the AI era and help businesses of all sizes grow together.
Customer Case: Fulike Logistics
Fulike is a leading brand in China's warehousing, logistics, and packaging e-commerce sectors. Its sales team must coordinate daily with customers, factories, finance, design, warehouses, and after-sales teams, resulting in an enormous communication burden.
Our solution is to create digital employees using Qwen Office, embedding them directly into collaboration environments (such as DingTalk IM) to handle routine tasks.
The Three Core Capabilities Supporting Digital Employees
First is Managed Agent: Users can customize the roles and permissions of digital employees in the backend, set system prompts and data connections, and the platform provides a stable operating environment.
Second is Enterprise Context: This ensures that digital employees possess proprietary corporate knowledge—not just being generic chatbots.
Third is Collaboration Integration: We've redesigned the bot access model, allowing Agents to join organizations as "employees," collaborating with human colleagues in the same workspace with appropriate permissions.
These three capabilities work together, enabling digital employees to truly integrate into business processes and achieve closed-loop collaboration.
A New Model for Complex Project Collaboration
In traditional workflows, project information is scattered across multiple groups, documents, and calendars, making progress synchronization difficult.
Qwen Office consolidates all relevant data, allowing humans and Agents to collaborate on a single interface, automatically summarizing information and executing decisions.
This feature is currently in gray-scale testing with select clients and will soon roll out widely. The goal is to center collaboration so that complex tasks can be completed end-to-end on a unified platform.
Real-Time Sharing of Frontline Experiences
In the past, outstanding cases proposed by top salespeople might take weeks to compile into scripts usable by the whole team.
Now, with the QwenNote A2 device, sales reps can simply say, "Generate a brief from our recent conversation," after visiting a client, and it automatically produces shareable knowledge assets.
QwenNote A2: Truly AI, No Recording
We've redefined the AI note-taking device. Unlike the previous generation's recording card, QwenNote A2 does not record audio by default.
After cloud-based speech-to-text (ASR) processing, the raw audio is immediately deleted. A small green light on the device indicates that it's used solely for note-taking, with no recording taking place.
This design balances privacy protection with efficient knowledge capture. We believe AI note-taking will become an essential tool in future workplaces.
Personalized Smart Workbench
Different industries and roles have distinct working habits. Qwen Office offers pre-configured workbenches tailored to sectors like education, finance, and law, as well as functions such as finance, HR, and administration.
It also allows users to create their own "Qwen Office apps," enabling highly personalized setups.
Frequently used features are arranged on the home screen according to personal preference, eliminating navigation and providing instant access to all information.
Safety: The Foundation of Trust
Meta once faced controversy after an AI mistakenly deleted emails. At Qwen Office, every action remains under control.
We establish boundaries beforehand, execute tasks in sandbox environments, require manual approval for critical actions, and implement post-execution safeguards to undo mistakes if necessary.
We ensure that every step of AI operation is supervised, giving users peace of mind.
All-Round Data Security Protection
Enterprise Context integrates vast amounts of sensitive data, so access must be tightly controlled.
Qwen Office requires user confirmation before accessing any sensitive information, and administrators can finely tune AI's data access permissions.
We also analyze AI-generated content to prevent accidental leakage of confidential information.
Security Center: Unified Management Platform
The Security Center integrates identity verification, permission management, and data access controls.
Operations require approval, protections leave logs, and risks are visible at a glance.
Through sandbox technology, risk management, encrypted transmission, and device security factors, we've built a multi-layered defense network.
For example, the enterprise version of Qwen Office only permits login on trusted devices, preventing data from ending up on mobile phones or personal computers.
Strong On-Premise Deployment Capability
Backed by Alibaba's self-developed chips and T-Head computing power, performance is tripled. Paired with Qwen's high-information-density models, we offer the most powerful private enterprise-grade office agent solution available in China today.
Whether it's the high-performance Max series or cost-effective options like Flash, Plus, and 27B Dense, all support on-premise deployment, meeting diverse security and performance needs.
Choosing Qwen Office Means Choosing a Digital Partner Who Understands Your Business, Can Collaborate, and Is Trusted
Real Customer Success Stories
As a Global Fortune 500 automotive company, GAC Group used Qwen Office to rapidly build secure diagnostic and refined operations systems, achieving in-house app development and automated process loops.
Transfar Group, a leading private chemical company, leveraged AI to assist in yellow phosphorus price analysis, optimizing raw material procurement strategies.
Jack Technology has topped global industrial sewing machine sales for 14 consecutive years and now uses Qwen Office to build an intelligent production and sales system, enabling traceable deliveries and predictable demand forecasting.