We have built a distributed consensus supercomputer that runs on top of IPFS. It can compute relevance for web3 data via the use of cyberlinks, that are created from IPFS hashes.
It is a distributed machine for answers (a relevance machine), built with the help of Cosmos-SDK/Tendermint/CUDA.
Within the network, web3-agents generate knowledge graphs with the use of what we call cyberlinks. A cyberlink is simply a link between two content-based links. This allows for a naturally semantic link, which is needed for the computation of relevance of subjects and objects within the knowledge graph.
IPFS provides significant benefits with regards to resource consumption. Cyber links and IPFS provide us with the superpowers that were inaccessible to previous architectures of general-purpose search engines.