Vector index for AI-powered actions and AI Assistant

Introduced in 2025

Updated in 2025.2

AI-Powered Insights and AI Assistant generate more relevant and helpful responses when the AI has access to your organization's internal knowledge, such as tickets and Knowledge Base articles. To make this happen, Alloy Navigator Express uses a vector index. This article explains how the vector index works, what content is included, how to enable it, and what to consider regarding data privacy.

What is a vector index?

A vector index is a specialized database that leverages semantic search to interpret and retrieve content by meaning rather than simple keywords. With this approach, AI-Powered Insights reliably delivers the most accurate and relevant information, despite variations in how your internal content is written or structured.

The underlying technology is based on a method called Retrieval-Augmented Generation (RAG). Here’s how it works:

1. The system converts your internal content into embeddings, which are numerical representations of meaning.

2. These embeddings are stored in a vector index.

3. When a user interacts with AI-Powered Insights, the AI searches this index for relevant information.

4. The retrieved content is used to generate a more informed response.

How AI-powered actions use vector index

Use of the vector index is controlled by the Allow AI to access internal knowledge option, available in the AI Integration > AI Configuration section. Once this option is enabled, the system begins indexing all historical content, creating and maintaining a searchable vector database. AI-Powered Insights (specifically, Suggest Solution / Next Steps / Action Plan and Analyze Risk) and the AI Assistant then query this index to deliver more accurate, context-aware recommendations by referencing your organization’s past tickets and knowledge base.

INFO: For details, see Vector index usage.

When the Allow AI to access internal knowledge option is disabled, AI-Powered Insights and the AI Assistant rely solely on the general system context and the current ticket; your historical content is not indexed or referenced. In this mode, each AI-Powered Insights' response will include a note recommending that you enable the vector index to leverage internal knowledge for more accurate, context-aware guidance.

When internal knowledge access is enabled

When you enable Allow AI to access internal knowledge in AI Integration > AI Configuration, the system automatically starts the scheduled task "AI-Powered Insights Index Maintenance".

This schedules task:

• Scans all existing tickets and Knowledge Base articles

• Converts them into embeddings

• Stores these embeddings in a vector index used for semantic search

• Runs every hour to keep the index up to date

The initial index build may take a significant amount of time and requires disk space. In the web-based Admin Center, you can monitor the indexing progress in the AI Integration > AI Configuration section, under Vector indexing progress:

• Last run: Date and time of the last job session

• Status: Last session status: Running, Success, or Failed

• Indexed: Number of items indexed vs. total (with percentage)

• Free space: Remaining space on the disk where the index is stored

• Database: Path to the index database (on the web server hosting your Alloy Navigator Express Web App)

To view or adjust the task, navigate to the Services and Integrations > Scheduled Tasks section, locate "AI-Powered Insights Index Maintenance" (#3149), and modify its schedule, view detailed logs, or temporary disable it.

Vector index and data privacy

The vector index contains only semantic vectors and does not store or share any readable content. However, whenever the index is built or updated, data from your tickets and Knowledge Base articles is temporarily sent to the AI provider specified in AI Integration > AI Configuration to generate those vectors. When a user triggers an AI-Powered Insight action or a customer asks the AI Assistant, and the vector index points to relevant content, that content, along with information from the current request, is sent to the AI provider as part of the prompt.

All data sent to your AI provider, including tickets, Knowledge Base content, and request details, is processed solely by the chosen AI provider under their terms of use, privacy policy, and data handling practices. Alloy Navigator Express does not store, analyze, or control information transmitted to the AI service. Please direct any questions on security, privacy, or compliance to your AI provider, and review their documentation to ensure it meets your organization’s requirements.

Related Information:

Configuring integration with AI

Supported OpenAI models for AI integration