Kingsoft Office is testing a more ambitious role for the business spreadsheet. Instead of asking an employee to choose among prebuilt templates or write code, WPS now proposes that a user can describe a need in ordinary language and receive a working business application inside the company’s existing table environment. The feature, called Inspiration Apps, appeared at the WPS AI NEXT media briefing in Beijing on August 28.
36Kr reported on August 29 that Inspiration Apps is part of WPS Multi-dimensional Tables. The company says a user can state a business requirement in natural language and AI will create a usable, personalized application that can continue to be refined. WPS is not presenting the feature as a clean-slate software generator. Its central claim is that the output remains connected to the customer’s existing data, permissions and collaboration functions in the table platform.
That design addresses a practical obstacle for workplace AI. A generated tool is of little value if it cannot see approved data, follow organizational roles or fit the process a team already uses. WPS says the resulting application can trace its data definitions and adjust its workflow, features intended to make natural-language creation more usable in a business setting rather than a one-off demonstration.
Inspiration Apps Bring Natural Language Into WPS Tables
The announcement places Inspiration Apps inside a familiar enterprise product rather than launching it as a stand-alone chatbot. WPS Multi-dimensional Tables already provides a setting in which teams organize information, assign access and collaborate. The new feature uses those elements as the foundation for AI-created applications. A user supplies a description of the work to be done; WPS says the system can turn that description into a tool that operates on the data and permissions already in place.
The choice of product context is important. Enterprise users typically do not want an AI system to produce a prototype that must then be rebuilt by an IT department before it can use real information. They need a process that preserves who can see a record, how an approval is handled and which field definition a team has agreed to use. WPS’s claim that the output can remain traceable and adjustable is therefore as important as the natural-language interface itself.
At the briefing, WPS described the feature as a way for personnel without programming expertise to build their own applications. This does not mean that every employee can replace a full software-development team. It means that some structured operational tasks may be expressed as a request, converted into an initial application and revised by the person who understands the workflow. The commercial value will depend on whether the system produces dependable outputs and whether organizations can govern them.
China’s workplace-agent market is becoming crowded with products that promise to turn a general-purpose model into an assistant for documents, screens and business processes. Alibaba’s QwenWork international expansion is one example. WPS is taking a more specific route: it is building AI generation around its own data-table product rather than first asking users to move their work into a separate agent interface.
Data Permissions Are the Hard Part of Workplace AI
The company’s description points to three requirements for a business-facing AI system: understanding the user’s business, maintaining a continuous data flow and remaining robust enough for a real process. Those requirements separate a workplace tool from an ordinary prompt. A sales manager, an operations planner or an education administrator may understand the business rule they want to automate, but they also need the generated system to use the right records and to avoid changing an approval path without notice.
WPS says the feature grows on top of existing data, permissions and collaboration capabilities. This approach could allow a company to reuse information already maintained in its tables. It could also make errors more consequential. If AI generates a workflow that treats a data field incorrectly or exposes a record to the wrong role, the problem is not just a bad piece of text. It is an operational and governance failure. The company’s emphasis on traceability and adjustable processes should therefore be treated as a product requirement that customers need to test, not a guarantee that every application will be safe.
The announcement gives two company-reported examples. Hudong-Zhonghua Shipbuilding used the product to create a lean equipment-management analysis system, while the Changping District Education Bureau used it to automate teacher-qualification review work. Those cases illustrate why WPS is emphasizing a table-based platform. Both involve structured records, repeated steps and staff who know the rules of the process even if they are not software developers.
WPS is gradually opening the feature to more users. That staged release leaves important questions unanswered: which types of business applications it can generate reliably, how a generated workflow is reviewed before deployment and how the company will handle sensitive internal data. But it also gives WPS a chance to observe how users describe work in practice, an input that may matter as much as the model behind the interface.
Kingsoft Competes Through Workflow Integration
The WPS launch shows how China’s AI competition is moving from headline model releases into product integration. Alibaba’s Qwen UI-Agent demonstrated a model designed to interact with screens and devices. WPS is applying a related ambition inside a workplace information system: transform a natural-language request into an action that fits a particular set of records and permissions.
That makes WPS’s advantage, if it has one, less about claiming the strongest general model and more about the installed context around its office software. Data tables, access settings and existing collaboration patterns can become a distribution channel for AI features. The challenge is that enterprise users will judge the result by accuracy, security and maintainability, not by whether the first generated screen looks convincing.
For Chinese small and medium-sized businesses, an AI function that can create a narrow internal tool without a large development project could be useful. The customer still needs to decide who owns the output, who validates it and how it changes over time. WPS’s assertion that workflows can be adjusted acknowledges that these applications are likely to evolve as users discover exceptions and new needs.
The August 28 unveiling is an initial product step, not evidence of broad adoption. WPS has not disclosed a user count or revenue for Inspiration Apps. The announcement nevertheless identifies a concrete direction for Chinese enterprise AI. Rather than treating AI as a separate conversation window, Kingsoft is attempting to place it inside the system where teams already organize the information that makes a business process possible.