Rui AI

Prerequisites

Feature Menu: Access to the AI Studio / Rui AI menu is required.

Operation Permission: Read and write access to AI Studio is required.

Data Permission: Access to at least one resource domain under an environment is required.

Model Configuration: At least one Large Language Model (LLM) must be added in the model list. If no model is configured, Rui AI will prompt you to add one.

Overview

Rui AI is the unified intelligent conversational interface of Bonree ONE AI Studio, integrating core capabilities such as Agent invocation and Skill execution.

Users can interact with Rui directly in natural language. A routing agent automatically identifies user intent and dispatches it to the most appropriate processing logic. Users can also proactively select a specific Agent or Skill to trigger a target capability precisely. Rui comes with a rich set of built-in general-purpose Agents that are ready to use out of the box, covering scenarios such as daily Q&A, fault diagnosis, data observation, and PromQL generation. It also supports invoking user-created personal Agents or Skills to meet customized business needs.

The interface supports three display modes — initialization, single-column conversation, and left-right split-screen — which switch automatically based on task complexity or can be toggled manually by the user, balancing the lightweight experience of simple Q&A with the structured presentation needed for complex diagnostic tasks.

Value

Unified entry point, lower barrier to adoption

Rui AI consolidates all AI Studio capabilities into a single conversational interface. Users can initiate tasks in natural language without needing to understand the underlying Agent configurations, significantly lowering the barrier to using AI capabilities.

Automatic intent recognition and intelligent task dispatching

When no Agent is specified, the routing Agent automatically parses user intent and selects the most suitable Agent or knowledge source to handle the request, reducing manual configuration steps and improving response efficiency.

Out-of-the-box, with support for deep customization

Multiple built-in general-purpose Agents are available with no configuration required. Users can also create personal Agents by uploading a dedicated knowledge base, connecting MCP tools, and configuring Skills — building an AI partner that truly understands their business.

Split-screen view for clear presentation of complex tasks

For structured output scenarios such as fault diagnosis and summary reports, the split-screen mode displays the conversation flow alongside the result report side by side, making it easy to follow up with questions while reviewing the full diagnostic conclusion, improving troubleshooting efficiency.

Multiple entry points, seamlessly integrated into operations workflows

In addition to the main AI Studio entry point, the fault diagnosis capability in Rui AI can also be invoked directly from the Alert List, enabling seamless integration between alert detection and AI analysis, and embedding AI capabilities into day-to-day operations workflows.

Use Cases

Daily Q&A and Technical Consultation

When operations engineers encounter technical questions in their daily work, they can ask directly in the Rui AI input box and receive immediate answers — for example, querying PromQL syntax, understanding the meaning of an alert metric, or getting guidance on feature usage. Rui responds quickly through its knowledge Q&A capability, eliminating the need to switch to a search engine or documentation system.

Emergency Fault Diagnosis

When a production environment alert is triggered, on-call engineers can invoke Rui with one click from the Alert List, or enter an alert ID directly in Rui AI to initiate a diagnosis. Rui automatically invokes the fault diagnosis Agent, performs multi-turn reasoning while calling tools such as monitoring data, logs, and slow queries, and ultimately outputs a structured diagnosis report containing fault clues, key evidence, and actionable remediation recommendations.

Skill Task Execution

After completing a diagnosis or analysis, operations personnel can select a custom Skill to quickly perform follow-up actions — such as sending DingTalk or WeCom notifications, creating tickets in internal systems, or pushing operations logs — achieving an end-to-end closed loop from analysis to execution, and reducing the manual effort of cross-system operations.

Personalized Agent Invocation

Users can connect their self-built personal Agents to Rui AI and invoke them directly from the conversational interface, meeting customized requirements for specific business scenarios — such as targeted diagnosis for a specific database type or performance analysis for a specific business system.

Operation Scenarios

Direct inquiry: Type a natural language question in the input box. If no Agent is selected, the routing Agent automatically identifies the intent and dispatches the request to the appropriate processing logic.

Query with a selected Agent: Click the "Agent" button in the toolbar below the input box to expand the Agent list, select the target Agent, then enter your question. This directly invokes the specified Agent, bypassing intent recognition.

Execute a Skill: Click the "Skill" button in the toolbar, select the Skill to execute, enter the relevant parameters, and send the request to trigger the Skill action.

Start a new session: Click the "New Session" icon in the top toolbar to clear the current conversation and return to the initialization screen to start a new session. Sessions are saved for 7 days and can be viewed and restored from the history list on the left.

Switch model: Click the model selection button in the bottom toolbar to switch the Large Language Model used for the current session (the model must already be added to the model list).

Getting Started

Accessing Rui AI

Log in to Bonree ONE.

Select AI Studio > Rui AI from the left navigation.

The initialization screen appears. You can type a question directly in the input box, or select an Agent / Skill before starting a conversation.

Starting a Conversation

Type your question in the input box and click the send button to submit.

If no Agent is specified, the system automatically identifies intent and routes the request to the appropriate processing logic.

To specify an Agent or Skill, click the corresponding button in the bottom toolbar to make a selection before submitting.

Viewing a Diagnosis Report (Split-Screen Mode)

When a fault diagnosis question is submitted, split-screen mode is activated automatically: the left side shows the conversation flow, and the right side displays the structured diagnosis report.

The report contains fault clues, key evidence, and remediation recommendations. Click "Export" to save the report. You can continue asking follow-up questions in the left input box.

Managing Session History

Click the "New Session" icon at the top to clear the current conversation and return to the initialization screen.

The session history list is located in the left panel (visible when expanded) and stores records from the past 7 days.

Click a history entry to restore that session's context and continue the conversation.