This recipe focuses on bringing help into the portal rather than sending users out to look for it.
Portals and dashboards in regulated industries concentrate a lot of information and actions. Users must interpret metrics and complete tasks while staying compliant. An in-portal assistant answers questions and guides steps right where users are working.
Build an AI Skill for “In-portal help” that takes current_screen, user_role, and intent as its main variables. The integration layer can set current_screen automatically based on the route or component, while the skill infers intent from the user’s question and only asks clarifying follow-ups when needed.
Ground answers with flat Topics such as Portal-navigation, Dashboard-metrics, Workflow-steps, and Configuration-help. Tag the relevant parts of your documentation and internal notes with these Topics so the assistant can respond with screen-specific guidance.
In some cases, the skill can trigger limited actions, like opening a related settings page or starting a predefined workflow. When it cannot complete a task safely, it should create a support ticket or internal task containing current_screen, user_role, the user’s question, and a short AI summary. This saves users from explaining everything again when humans step in.
On the analytics side, you can aggregate questions by current_screen to see where confusion is highest. The assistant can send regular notifications summarizing “top 5 screens by help requests” and the most frequent topics, giving product and CX teams concrete areas to improve.
Within the portal, you can also show in-app notifications that depend on user_role and simple behavioral segments. For example, new admins might see onboarding hints, while experienced users see reminders about advanced features they have not touched yet. These are powered by the same segmentation layer that the AI Skills use.
Conclusion
An in-portal task and journey guidance assistant brings context-aware help into the heart of your product. AI Skills and flat Topics keep answers precise, while internal tasks and notifications turn user questions into design insight. Segmented in-app notifications complement the assistant, creating a more supportive and efficient portal experience.