AWS announced a generative AI-based solution to modernize and scale support operations by automating the creation of Standard Operating Procedures (SOPs) from training videos. The system uses Retrieval Augmented Generation (RAG) to guide ticket resolution and machine learning to optimize workload distribution. This approach aims to reduce the time analysts spend searching for guidance, allowing them to focus on resolving issues more efficiently. The solution also automates operational tasks such as ticket tagging, commenting, and status updates through agentic workflows, while maintaining human oversight for accuracy. Source: awsml
Support operations face challenges such as fragmented knowledge across documentation, people, and tools, leading to inefficiencies and inconsistent quality. Teams maintain hundreds or thousands of SOPs, but these are often developed on an as-needed basis without a systems perspective. As a result, documentation rarely reflects end-to-end processes, and updates are made locally, creating a gap between what is documented and what is real. Knowledge is also shared through training calls and recordings, but this information is not easily accessible after the sessions end. The solution addresses these issues by capturing knowledge directly from operational activity and preserving it in a structured, searchable format. Source: awsml
The solution also tackles the problem of tickets arriving faster than guidance can be found, which leads to backlog and delays. Analysts often spend significant time searching for SOPs across various sources, and even when the right document is found, it typically covers only one function’s view of the process. This forces analysts to mentally stitch together multiple documents, tribal knowledge, and past experience to reconstruct the full resolution path. The system aims to automatically identify guidance from the content of the request itself, reducing the reliance on manual searches and improving resolution times. Source: awsml
