When we use LangChain to interact with AstraDB, there's a disconnect in the visual representation of the flow. For example, in this diagram, it appears that the LangChain chatbot is independent from the AstraDatasource.
(Source is the new Azure document ingestion demo)
The true representation of the flow is that the docs ingested by the top pipeline are used by the chatbot.
So, in this case, the diagram above is misleading because it appears like the PDF ingestion flow is unrelated to the chatbot.
It would be a big improvement to the usefulness of the UI if we could find a way to fill gaps like this.
When we use LangChain to interact with AstraDB, there's a disconnect in the visual representation of the flow. For example, in this diagram, it appears that the LangChain chatbot is independent from the AstraDatasource.
(Source is the new Azure document ingestion demo) The true representation of the flow is that the docs ingested by the top pipeline are used by the chatbot. So, in this case, the diagram above is misleading because it appears like the PDF ingestion flow is unrelated to the chatbot.
It would be a big improvement to the usefulness of the UI if we could find a way to fill gaps like this.