implement a semantic similarity threshold to determine whether:
the agent would simply just be a chatbot with personality,
or if it should use its tools to its advantage for more sophisticated questions.
This may require constant querying of the VDB
For example,
UI: Hello, how are you?
R: (Langchain chaining) I am good, how are you today?
UI: What are vector databases?
R: (Threshold > 0.85 to Pinecone DB) (Answer)
Either this, or have it utilize its toolbelt to determine whether it should VDB retrieval and RAG, search the web, or respond as a chatbot.
implement a semantic similarity threshold to determine whether:
the agent would simply just be a chatbot with personality,
or if it should use its tools to its advantage for more sophisticated questions.
This may require constant querying of the VDB
For example,
UI: Hello, how are you?
R: (Langchain chaining) I am good, how are you today?
UI: What are vector databases?
R: (Threshold > 0.85 to Pinecone DB) (Answer)
Either this, or have it utilize its toolbelt to determine whether it should VDB retrieval and RAG, search the web, or respond as a chatbot.