Perplexity is one of the only tools that I think takes great advantage of LLMs strengths (pattern matching, interpolation) while also making up for its weaknesses (hallucinations, lack of ability for novel contribution/thought) by using it for summarisation with search.