How elvex Categorizes and Tracks Requests
elvex tracks usage through a comprehensive request categorization system that captures different types of user interactions. Understanding these categories helps explain why your total numbers might not always match what you expect.
Request Type Definitions
Total Requests
The sum of all interactions with elvex across all channels and methods. This includes conversations, API calls, and system-generated requests. Think of this as the complete picture of your organization's elvex activity.
Home Assistant Conversations
Requests made through the elvex home chat interface. These are conversations where users interact directly with elvex's general-purpose assistant without using a specific custom assistant. Each message exchange in the home chat counts as one request.
Assistant Conversation Requests
Interactions with custom assistants that your organization has created. This includes both direct conversations with assistants and any background processing they perform (such as accessing datasources or running tools). Each message to a custom assistant generates at least one request, but complex assistant behaviors may generate additional requests.
API Requests
Direct calls to the elvex API, typically from integrations, custom applications, or automated systems. These bypass the web interface entirely and represent programmatic usage of elvex.
System Requests
Background operations initiated by elvex itself, such as data processing, maintenance tasks, or automated assistant behaviors. These requests happen without direct user interaction but are necessary for elvex functionality.
Slack Requests
Interactions that occur through the elvex Slack integration. These are tracked separately because Slack conversations have different characteristics than web-based interactions.
Why Request Numbers Might Not Add Up
The relationship between these metrics is more complex than simple addition. Here's why your numbers might not align:
Request Multiplication
A single user action can generate multiple requests. For example:
An assistant conversation that accesses a datasource creates both an assistant conversation request and potentially system requests for data retrieval
API calls that trigger assistant responses generate both API requests and assistant conversation requests
Complex assistant workflows can cascade into multiple system requests
Overlapping Categories
Some requests belong to multiple categories simultaneously. A Slack message to a custom assistant generates both a Slack request and an assistant conversation request.
Background Processing
System requests often occur without corresponding user-visible actions, which can make the total requests number appear higher than expected based on visible user activity.
Privacy and Access Control in Analytics
elvex's privacy model directly impacts what data you can access through the analytics page interface. It does not impact the numbers you see, just your ability to drill down into specific details.
Private Assistant and Datasource Limitations
Private Assistants
If an assistant is marked as private, you cannot click through to view its details from the analytics page unless you have explicit access to that assistant. The usage data (request counts) will still appear in your aggregated statistics.
Private Datasources
Similarly, private datasources contribute to overall usage metrics but their specific details are not accessible through analytics unless you have permission to view them.
Data Visibility Principles
elvex follows a principle of aggregated transparency with granular privacy. This means:
Overall usage patterns and numbers are visible to administrators
Specific content, conversations, and private resource details remain protected
Users maintain control over their private assistants and datasources even in analytics contexts
Understanding User Metrics
Active vs. Total Users
The analytics page focuses on active usage rather than total enrollment. This design choice emphasizes engagement and actual platform utilization over simple user counts.
β
The total number of users in your organization is not directly displayed on the main analytics dashboard. The "Active Users" line in the chart shows users who made requests during the selected time period. To find your complete user count, set your time period to your desired time period and look at the total number displayed in the User table.
Mental Models for Analytics Success
Think in Terms of Workflows, Not Just Messages
A single user goal (like "analyze this data") might generate multiple requests as elvex accesses tools, datasources, and processes information. The total request count reflects the complete computational work, not just user-visible interactions.
Privacy as a Feature, Not a Limitation
The inability to access certain private resources through analytics is intentional design that maintains user privacy while still providing organizational insights. This protects sensitive information while enabling usage monitoring.
Analytics as Trends, Not Absolute Truth
Use analytics to understand patterns, peak usage times, and adoption trends rather than focusing on exact request counts. The value lies in understanding how your organization's elvex usage evolves over time.
Understanding these principles helps administrators make better decisions about resource allocation, user training, and elvex deployment strategies while respecting the privacy and security boundaries that make elvex suitable for enterprise use.