Marvis AI Assistant Overview

Marvis AI Assistant (Marvis) is an AI-enabled conversational interface that helps you understand, troubleshoot, and explore Juniper Networks products and services. It combines natural language interaction with specialized AI agents to deliver accurate, contextual, and actionable responses.

Marvis is designed to serve as your first-level assistant. It helps you diagnose issues, understand system and network behavior, and access relevant documentation through a single, intuitive interface.

Marvis uses natural language understanding to interpret your query and then orchestrates the appropriate AI agents to analyze data, retrieve relevant information, and generate responses.

By cross-referencing and collating information from multiple authoritative sources, it enables users to resolve issues faster without having to depend on multiple sources.

When you start a conversation with Marvis, it:

• Uses natural language understanding (NLU) to interpret intent and context

• Simultaneously, Marvis invokes purpose-built AI agents to:

  • Perform diagnostic analysis using available network telemetry, logs, and configuration data

  • Analyze application specific data such as operating system logs, blast radius, crash information

  • Correlate insights across the network to identify patterns, anomalies, or potential root causes

  • Retrieve relevant product documentation and provide contextual guidance

• Maintains conversational persistence across follow-up queries and sessions

The AI agents are invoked in response to a user request and operate within predefined scopes. This enables Marvis to deliver more accurate diagnostics, guided troubleshooting, and provide context-aware recommendations and documentation.

Marvis is designed to assist only with Juniper Networks products and services.

Marvis generates responses supported by citations to authoritative Juniper resources. Users can select citation links to view source materials directly. If a query is unrelated to Juniper Networks products or services, Marvis informs the user that it is designed exclusively to assist with Juniper-related topics and does not generate speculative or unsupported responses.

To enhance usability and effectiveness, Marvis also provides follow-up questions tailored to the current query, easy access to actions such as copying responses, viewing citations, and resetting conversations.

A sample scenario can look like this; you notice intermittent latency in a data center fabric and ask Marvis to find the root cause.

Marvis interprets your description of the issue using natural language understanding to determine the scope and intent of the query. Based on this understanding, it invokes specialized AI agents to analyze relevant network telemetry, logs, and recent configuration changes. It then correlates anomalies across the fabric and retrieve applicable Juniper documentation. And then presents you guided troubleshooting steps with supporting references and allows you to refine the investigation through follow-up questions within the same conversation.