Rajesh Cherukuri

Cohere reranking greatly improved the results of our RAG

Strongly recommend for anyone building a file uploader solution. Comes with a lot of cool features and easy to get off the ground

We use Pinecone as our Vector Database. Easy to get started, simple to use, always up and running.

We use LangChain to build our RAG pipeline.

We use the MarkMap open-source library after AI responses to convert completed answers into markdown format optimized for rendering fully interactive mind maps that can open in full screen mode This solves the problem of increasingly long AI responses that are hard to consume by providing visual hierarchy.