Article body
Full article
Partners add analytics to CRM, ERP, industry SaaS, and other business products so users can obtain traceable data while completing business tasks. Product teams need to deliver a question interface, result presentation, metric definitions, and access boundaries inside the workflows that users already follow.
HENGSHI’s View: Deliver Analytics with the Business Workflow
HENGSHI treats embedded analytics as a capability of the business system, not as a link to a separate page. When users ask questions from customer, order, inventory, or operating pages, the system needs to return results that fit the current business context and show the data scope and metric definitions behind them.
- Data questions: Give users an entry point on the business page for a continuous conversation or a one-time task.
- Result presentation: Place the chat window, result cards, charts, or dashboards on the page where users work.
- Metric semantics: Include business terms, metric definitions, and available data scopes in the solution design so the same term does not acquire different meanings across pages.
- Access boundaries: Use business identity, application resources, data packages, and row and column rules to control what each user can view and do.
Embedding a dashboard delivers a presentation layer. Turning analytics into a product capability also requires teams to connect these four elements with business objects, user identities, and operating flows.
Assign Runtime Responsibility Before Choosing an Integration Method
HENGSHI SENSE Data Agent integration separates the operating mode from the integration method. The operating mode answers one question: who keeps a data conversation moving? Agent mode preserves the interactive experience of continuous execution, state changes, tool calls, and chart results. In Workflow mode, a backend workflow service performs the work while the frontend starts the task, reads its state, and presents the result.
The integration method determines how the capability enters the product. iframe, JS SDK, and API are integration methods. They do not define whether an Agent or Workflow owns execution.
| Integration method | Suitable responsibility boundary | What the partner receives |
|---|---|---|
| iframe | The partner wants to reuse a complete page with limited frontend work. | HENGSHI provides the chat interface, history, result cards, and chart presentation. HENGSHI SENSE also supports embedding at the platform, module, feature, application, or folder level. |
| JS SDK | The partner keeps its own page shell, entry points, and business interactions. | The partner can embed a complete chat window or the result card for one question. After the parent page loads the Copilot JS SDK, it can take control of the assistant entry point inside an iframe dashboard. |
| API | The partner already has backend orchestration, a multi-agent system, an AI gateway, or a workflow service. | The partner owns orchestration, page presentation, and business integration. HENGSHI provides data-question interfaces. |
Product owners should first decide whether a frontend Agent runtime or backend Workflow will keep the conversation moving. They can then choose between reusing a complete page, reusing frontend capabilities, or calling backend capabilities alone. This order shapes session ownership, page boundaries, authentication, and troubleshooting responsibility.
Bring Business Context into the Analysis
Embedded analytics should begin with a defined business action. A sales manager reviews account health on a customer page, a store manager reviews store performance, and a buyer handles an inventory exception. Each task requires a different entry point, question scope, and result destination. Partners should put the business object, user identity, and available data scope into the integration design instead of placing a general question box in a corner of the system.
In the JS SDK, userConfig can specify the data available to the current conversation. It supports configurations based on data packages and datasets or on subject areas. Partners need to map business pages to their data scopes and manage the names, calculations, and presentation context of key metrics. Users see a natural-language conversation. Product teams maintain the explainable data scope and business semantics behind it.
Treat Permissions as Part of the Product Capability
Once analytics enters a business system, access control must extend into the analytical experience. HENGSHI SENSE can query application and data permissions by user, user group, organizational structure, and tenant. Access management covers resources such as applications and data packages. Administrators can also query or edit row and column permissions for applications and data packages and inspect button-level privileges inside an application, including detail access and data export.
Partners need to map their business roles to these resource permissions, then test page visibility, data scope, detail access, and export actions with different roles. The design must cover both the login identity in the business system and the applications and data resources used by the analytical result.
What Partners Need to Deliver a Productized Experience
Partners can begin with one high-frequency business page and place the analytical entry point and results in the user’s current workflow. They should then assign the runtime responsibility to Agent or Workflow and the integration responsibility to iframe, JS SDK, or API. Launch acceptance should test how results, available metrics, and permitted actions differ by role in the same business scenario.
HENGSHI provides embeddable data questions, result rendering, and application and data-resource management. Partners own industry workflows, business objects, and user experience. This division of responsibility enables both sides to deliver analytics as a stable product capability.