I built DiffFind because traditional search and diff tools are great at finding exact text, but they struggle when the wording changes while the meaning stays the same.
DiffFind combines traditional lexical diffing with semantic search and semantic comparison for text and documents.
For Semantic Find, I use semantic retrieval followed by LLM reranking and extraction. In my testing, this has worked significantly better than simply giving a document to an LLM and asking it to search.
It currently supports text and PDF documents, plus a REST API for programmatic use.
I'd love feedback from HN, particularly on the semantic search/comparison approach and where you see it being useful.
I built DiffFind because traditional search and diff tools are great at finding exact text, but they struggle when the wording changes while the meaning stays the same.
DiffFind combines traditional lexical diffing with semantic search and semantic comparison for text and documents.
For Semantic Find, I use semantic retrieval followed by LLM reranking and extraction. In my testing, this has worked significantly better than simply giving a document to an LLM and asking it to search.
It currently supports text and PDF documents, plus a REST API for programmatic use.
I'd love feedback from HN, particularly on the semantic search/comparison approach and where you see it being useful.