[UI copy]: AI Components Copy

[0lha0liinyk]

Description

Background & Context

We’re introducing an AI Summary component in flyouts that will be reused across multiple workflows, including Entity Summary, Alert flyout, and Attack flyout.
The component supports several states: not generated, in progress, generated, regeneration failure, and error states.

Design and layout are already approved. What’s needed now is final UX copy for error states that is short, clear, consistent, and actionable, without exposing technical details or overwhelming users. The same copy will be reused across all flyouts.

Error Categories to Cover

We intentionally group multiple technical failure scenarios into two user-facing error types.

1. General AI generation error

This error groups:

Backend errors

• Internal server errors
• Upstream service failures
• Unexpected response formats from backend services

LLM-related issues

• Model processing failures
• Invalid or incomplete LLM responses
• Token limit exceeded
• LLM service downtime or overload

Suggested copy (draft):

Title: Error generating summary
Body:
Due to an unexpected issue, the summary could not be generated. Please try again shortly, or adjust the time range or AI model if the issue persists.

2. Connection or timeout issue

This error groups:

• No internet connection
• Unstable or slow network
• Service timeout
• API unreachable

Suggested copy (draft):

Title: Connection issue
Body:
Due to a connection issue or timeout, the summary could not be generated. Please check your network and try again.

Empty States – AI Priorities Component (Overview / Entity Analytics)

In addition to flyouts, the AI Priorities component will be used on the Detections Overview dashboard and in the AI Priorities panel for Entity Analytics.

This component needs clear, non-interactive messaging for multiple states where priorities cannot yet be shown or are being processed. Unlike flyouts, this component does not allow any manual user actions (no retry, no regenerate). The system will attempt to regenerate priorities automatically.

A key UX requirement is to clearly communicate that AI priorities are not affected by the time range selected in the KQL search or calendar, to avoid user confusion.

States to Cover

The AI Priorities component should support the following states and corresponding UX copy:

1. General AI generation error (Same category and behavior as in flyouts)
   Goal:
   Explain that priorities couldn’t be generated due to a system issue, reassure the user, and indicate automatic retry.

2. Connection or timeout issue (Same category and behavior as in flyouts)
   Goal:
   Clearly indicate a temporary connectivity issue without asking the user to do anything.

3. Not enough data to generate priorities
   Occurs when there is insufficient signal, events, or correlations to produce meaningful priorities.
   This is not an error and should be communicated neutrally.

Goal:
Explain why priorities are unavailable without implying a failure.

4. Generation in progress
   Priorities are currently being generated. The system is working automatically in the background.

5. Personalized priorities generated
   AI-generated priorities are available and displayed
   Copy should reinforce that these priorities are global and not tied to the selected time range
   In addition we need to improve the button copy.

For example, for the “Personalized priorities generated” state, when AI detects multiple priorities (e.g., 10 total), only a subset should be shown directly in the panel (for example, the top 4). These and remaining priorities should be accessible via the AI assistant. The button text should clearly communicate this behavior.

Related links / assets

Figma link(s): https://www.figma.com/design/ccUbKmELpHVGJeiz3kLXzw/-9.3---Entity-Analytics-Portfolio?node-id=10153-93729&t=nZv46s9bkIJuAnyI-1

Which product area does this mainly concern?

Cloud UI

Collaborators

Olha Oliinyk

Timeline / deliverables

We need the final copy in the nearest possible term.