According to CoinMeta, Datadog has launched its first dedicated operations and maintenance model, Scout 1.0, which utilizes the open-source GLM-5.3 Flash from Zhispu. This model is specifically designed to troubleshoot online system failures. It can query logs, monitor indicators, and service call chains to gradually identify the root causes of problems. Datadog builds a simulation environment using real accident data and monitoring information, and trains the model through reinforcement learning to enable it to identify error clues and support diagnostic conclusions with evidence. In the 90 fault troubleshooting tasks of the public benchmark SRegym, Scout completed 80 diagnoses and repairs, tying for the highest score with GPT-6 Astra Xhigh. It only fell short of GPT-6.1 SOL Xhigh by 2 tasks. Scout has been integrated into Datadog's Bits Investigation, and at present, it is only available in a closed preview for a limited number of customers. The main focus at this stage is on identifying the causes of faults and providing recovery suggestions; automatic execution of repairs remains a goal for future development.