Hey there, I'm currently struggeling with adding a FAQ-first feature to my autonomous nodes. The idea is to - Always search the FAQ knowledge base for things that could slightly affect the response - Always search a second knowledge base that has the "real" data I need them to be seperate, because there might be FAQs like Q:"when doese the skiing resort open?" A:"skiing resort will open on the 04.12.2025" And somebody asks for the lift services. Or a QA about a blocked road or something similar. I tried adding 2 search knowledge cards with the different KBs but the LLM never uses both tools, however hard I try with the prompts (also card prompts). Is there a best practice on how to achieve that?