We are now at the superhuman AI phase! @eyelevelai's GroundX, and enterprise-grade RAG system, outperformed humans on DocBench — a benchmark that tests deep document comprehension. GroundX is an open-source system that you can run on your servers (or any cloud provider, as long as you have access to GPUs) and works without a network. (If the military wants to do RAG, this is precisely what they will be looking for.) They offer two services you can use: 1. Ingest: This service uses a pretrained vision model to ingest and understand your knowledge base. 2. Search: This service combines text and vector search with a fine-tuned re-ranker model to retrieve information from your knowledge base. The combination of these services can read and understand documents better than humans! (Here, we're talking about legal documents, medical records, and financial reports.) While your average AI still fumbles tables and misreads figures, GroundX crushes multimodal and textual questions with 90–95% accuracy. That's superhuman level on one of the hardest document tasks out there! And this is not just about extracting text, but about understanding the structure of a document, its visuals, context, and nuance. The funny thing is that GroundX has almost saturated the benchmark. We'll need better benchmarks if we want to keep measuring progress. This is huge for RAG systems. Remember that RAG isn't just about finding the right chunk of text. RAG is about delivering high-quality, context-aware answers from complex documents. GroundX's progress makes it really hard to beat it at this game.