Does anyone know how to build a search

I'm pretty sure Perplexity just uses Bing Search under the hood. They probably do some other processing but it's mostly just a search API connected to a LLM.

Could probably build it with this: https://llamahub.ai/l/tools/llama-index-tools-bing-search?from=

But how do they synthesize answers so well? It can't just be RAG, there must be something else

I was watching this yesterday: https://youtu.be/yGejxO1xYmo?si=22UtE4T0RVXbqYOy

he mentions that there are different operations for different items, like there are multiple llm calls just to provide the final rich response.

ah nvm nvm