Continuous profiling has emerged as a crucial component of observability in software development, complementing logs, metrics, and tracing by providing insights into CPU usage and performance inefficiencies in native codebases. Initially a practice mainly for managed languages, recent advancements, particularly in eBPF technology, have made continuous profiling applicable to native code as well. This approach allows developers to detect performance issues proactively, conduct differential performance analysis, achieve cost savings in cloud deployments, and identify inefficiencies without preconceived notions, thereby addressing "unknown unknowns" in application performance. The blog post illustrates the practical application of continuous profiling using tools like Grafana Alloy and Grafana Cloud Profiles in benchmarking open-source applications like DuckDB and PostgreSQL, demonstrating minimal performance impact and providing actionable insights for CPU optimization. The availability of off-the-shelf components has significantly simplified the process of gaining detailed performance insights, making it accessible to development teams without the need for specialized engineering skills.