Performance Profiler

What it does

The OneWave AI Performance Profiler skill analyzes and visualizes the performance of AI agent interactions. It provides detailed metrics such as token usage, latency, and cost breakdown for each turn in a conversation. This allows users to understand resource consumption patterns and identify areas for optimization within their AI workflows.

When to use it

• Debugging slow responses: Identify bottlenecks causing delays in an AI agent's replies.
• Cost Optimization: Analyze token usage per interaction to reduce operational expenses.
• Performance comparisons: Evaluate the efficiency of different prompts or model configurations.
• Understanding resource consumption: Gain insight into how your AI agents are utilizing resources over time.

Key capabilities

• Token Usage Tracking: Monitors and reports on the number of tokens used in each turn.
• Latency Measurement: Records the time taken for responses to be generated.
• Cost Breakdown: Provides a detailed cost analysis per interaction.
• Visualization: Presents performance data in an easy-to-understand format.

Example prompts

• "Analyze my last conversation with the AI agent and show me token usage."
• "What was the average latency for the previous 10 interactions?"
• "Show me a cost breakdown of my recent AI agent use."

Tips & gotchas

The skill requires access to your AI agent interaction history. Ensure that the necessary permissions are granted for accurate data collection and analysis.