Tracing a GenAI App
MLflow Tracing captures your GenAI app's execution flow, providing visibility into every step from user input to final output. See exactly what happens inside your application - including prompts, model calls, tool usage, latencies, and token counts.
In this quickstart, you'll build a simple GenAI app that answers questions while automatically capturing detailed traces for debugging and optimization.
Choose the quickstart guide based on your preferred development environment:
- Locally in an IDE or notebook - Use any local development environment such as an IDE (VS Code, PyCharm, Cursor, etc) or a locally-hosted notebook environment (Jupyter, etc)
- Databricks-hosted Notebook - Use a hosted Databricks Notebook
Next steps
Continue your journey with these recommended actions and tutorials.
- Evaluate your app's quality - Systematically test and improve your traced application
- Collect human feedback - Add developer annotations and gather expert insights
- Advanced tracing techniques - Learn automatic and manual tracing patterns
Reference guides
Explore detailed documentation for concepts and features mentioned in this guide.
- Tracing concepts - Understand the fundamentals of MLflow Tracing
- Tracing data model - Learn about traces, spans, and attributes
- Query traces - Explore how to programmatically access trace data