Part of what we're doing is building a platform that captures the broader lifecycle of tasks beyond "inference alone" -- things like data import, index building & maintenance, drift detection, corpus query.