### Answer --- The documentation doesn't explicitly state a latency difference between URL sources and text snippets in the knowledge base. [(1) ]() However, the documentation does recommend preferring

.md

(Markdown) files because well-structured Markdown is chunked and retrieved more accurately. For best retrieval performance, the documentation suggests: - Using clear, descriptive headings and keeping each section focused and reasonably short - Writing short paragraphs and lists to separate concepts - Grouping related information within the same section so chunks remain cohesive Both URL and text sources go through the same process during knowledge base creation: they are chunked, embedded, and stored in a vector database. During calls, the system retrieves the most relevant chunks regardless of the original source type. If latency is a concern, you can also adjust the "Chunks to retrieve" setting (range 1-10, default 3) and "Similarity Threshold" (default 0.6) to optimize retrieval.

​ To learn more, see the sources I considered: [[1] Knowledge Base](https://docs.retellai.com/build/knowledge-base) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.