Is there a way for llama index, instead

I think if it was an agent, it could do that. Or some kind of custom query engine + router/sub question query engine

(cc @emrgnt_cmplxty -- this is exactly the type of use-case a sciphi retriever might solve)

Do you think if I do an agent with a tool as a query engine along with a web retriever its possible?

I don't know how I would detect when the engine doesnt give context

I think it would have to either decide ahead of time based on tool descriptions

You could also implement a custom query engine that first calls a query engine, calls the LLM to decide if if "no context response" was given, and if so, call another service like a web retriever tool

Anything is possible with a bit of custom code 🙂

https://docs.llamaindex.ai/en/stable/examples/query_engine/custom_query_engine.html