The AI Assistant feature is currently in Beta.

AI Analysis lets you get AI-generated insights from a completed load test report. Instead of manually reviewing charts and raw session data, you can ask the AI Assistant direct questions about test results, and it will analyze the report data to answer them.

This article covers how to request AI Analysis on a Load Test Report and includes example prompts to get started.

Prerequisites

  • A completed load test with a generated report.
  • The AI Data Analysis account setting turned on. A Dotcom-Monitor account Administrator must enable this setting before the AI Assistant becomes available on any report. For activation steps, see AI Data Analysis Settings.

Working with AI Assistant

To run AI Assistant, open a load test report and select the AI Assistant tab. Review and accept the AI Data Processing Consent, Disclosure & Accuracy Notice, then click Start AI Analysis.

You can enter your own prompt or choose one of the predefined prompts available in the tab. AI Assistant analyzes the report data and displays its findings in the AI Assistant tab.

AI Assistant analyzes only the data available in the selected report. It does not have access to external information such as server logs, application code, or your project documentation. If additional context is needed for the analysis, include the relevant information in your prompt.

Example Prompts

Below are some additional prompt ideas. Use these prompts as a starting point. Adjust them to match your test goals.

Tip:  Specific prompts produce more specific answers. Instead of “How did the test go?”, ask about a particular metric, time range, or step.

  1. “Which step or request in the user flow is the biggest bottleneck?”
    Identifies the specific request or page step with the longest response time.
  2. “Is the failure rate caused by the application, the network, or the load injectors?”
    Helps distinguish an application-side issue from a test infrastructure limitation.
  3. “Did load injector CPU or memory usage limit the test, or was the target system the bottleneck?”
    Determines whether test results were constrained by the load injectors themselves rather than the system under test.
  4. “How quickly did response times recover after each load spike?”
    Evaluates system resilience by measuring recovery time between load peaks.
  5. “What is the most common error type, and which step in the flow triggers it most often?”
    Pinpoints the exact location in the user flow where a specific error is concentrated.