Hey Praveen - great question.

1A. Combining structured and unstructured data elements is exactly what we think there's a lot of value. Check out our write-up about our CSV to Graph feature where we have a case study that combines structured and unstructured data sources: https://medium.com/enterprise-rag/tables-csvs-to-graphs-whyhow-ai-sdk-9ed6d95f2f8b

1B. Agreed - we believe vector + KG hybrid search is the most optimal. We have this feature natively with our SDK. We write about this more here with our Chunk Linking feature: https://medium.com/enterprise-rag/whyhow-ai-kg-sdk-upgrade-vector-chunk-linking-with-graphs-increasing-explainability-accuracy-cc16c956ae42

2. Time series RAG is certainly possible with graphs, though more complex. You'd likely want to use something like putting the time data as a property, and then performing some agentic post-graph extraction processing