Filtering: Advanced run analysis with new filters and AI Query in LangSmith

Whether you’re trying to understand the experience of a specific customer segment or find traces associated with poor user feedback scores, filtering allows you to focus on the subset of traces that matter to you and to drill down into the details you need. LangSmith lets you filter based on a variety of trace attributes, and also offers full text search on inputs and outputs. We’ve added new functionality for allowing for filtering of runs based on attributes of either their *parent* or *child* runs. This is particularly useful for filtering runs that are part of an agent (or any multi-step) workflow. Documentation: https://docs.smith.langchain.com/monitoring/faq/filter Blog post: https://blog.langchain.dev/langsmith-production-logging-automations/