Sentry is using machine learning to improve issue creation and alerts by automatically grouping similar issues together, reducing false positives, and prioritizing issues based on their likelihood of being important. The new approach uses a powerful embeddings model to create a semantic representation of error stack traces, enabling efficient comparison and automatic grouping. This results in a 30-50% reduction in issue creation and fewer grouping mistakes. Additionally, Sentry is introducing an issue priority model that analyzes incoming errors and identifies low-priority issues, automatically tagging them as such and reducing noise in the alert stream. The goal is to make Sentry more actionable and easier to use by minimizing time spent on debugging and maximizing time spent on building great user experiences.