Analytics

Monitor queries, track costs, and analyze usage across your workspace.

Overview

The Analytics section provides visibility into how your agents are being used:

• Queries: View detailed logs of all agent queries
• Ingestion: Track document processing costs and status

Queries

The Queries page shows detailed logs of all queries made to your agents.

Summary Stats

At the top of the page, you'll see key metrics:

| Stat | Description | |------|-------------| | Total Queries | Number of queries in the selected time period | | Total Cost | Combined cost of all queries | | Avg Response Time | Average time to generate responses |

Filters

Filter the query log by:

• Time: Select a date range (Last 7 days, Last 30 days, Custom)
• Agent: Filter by specific agent
• User: Filter by user who made the query

Query Log Table

The table displays:

| Column | Description | |--------|-------------| | Date | When the query was made | | Query | The user's question (truncated) | | Agent | Which agent answered | | User | Who made the query | | Time | Response generation time | | Cost | Token cost for this query | | Citations | Number of document citations used | | Definitions | Number of logic definitions applied |

Click any row to view the full query details including:

• Complete query text
• Full response content
• All citations with excerpts
• Applied logic definitions
• Token breakdown (prompt, completion, total)

Exporting Data

Click Export to download query data as CSV for further analysis.

Ingestion

The Ingestion page tracks document processing across all datastores.

Summary Stats

| Stat | Description | |------|-------------| | Total Documents | Number of documents processed in the period | | Total Cost | Combined ingestion cost |

Filters

Filter the ingestion log by:

• Time: Select a date range
• Datastore: Filter by specific datastore
• Status: Filter by processing status (Completed, Failed, Processing)

Ingestion Log Table

| Column | Description | |--------|-------------| | Date | When the document was uploaded | | Document | Filename or document name | | Datastore | Which datastore it was uploaded to | | Cost | Processing cost for this document | | Status | Processing status |

Status Values

| Status | Description | |--------|-------------| | Completed | Successfully processed and indexed | | Processing | Currently being processed | | Failed | Processing failed (click for error details) | | Pending | Waiting in queue |

Understanding Costs

Query Costs

Query costs are calculated based on:

• Input tokens: Your query + retrieved context
• Output tokens: The generated response
• Retrieval: Document search operations

Ingestion Costs

Ingestion costs are calculated based on:

• Parsing: Extracting text from documents
• Chunking: Splitting into searchable segments
• Embedding: Generating vector embeddings
• Storage: Storing document content and embeddings

Best Practices

Monitoring Query Performance

1. Track response times: Identify slow queries that may need optimization
2. Review citation counts: Low citation counts may indicate retrieval issues
3. Check definition usage: Ensure logic definitions are being applied

Optimizing Costs

1. Right-size retrieval: Adjust Top K results to balance quality vs cost
2. Review failed ingestion: Fix issues to avoid re-processing
3. Monitor usage patterns: Identify opportunities for optimization

Next Steps

• Creating Agents - Optimize agent configuration
• Logic - Improve answer quality with definitions
• Team - Manage who has access to analytics