Quesma has launched OTelBench, an open-source suite to benchmark OpenTelemetry pipelines and AI-driven instrumentation. It evaluates collector performance under stress while testing how accurately LLMs handle complex SRE tasks like context propagation. Initial data shows AI agents often achieve success rates below 30%, highlighting the gap between code generation and production observability. By Mark Silvester