Hey HN. I have been building TESSRAL for a while. It is a deeptech product which lets you train your own image or voice AI models with a small amount of processing power and a small dataset belonging to you. Or you may use other TESSRAL users' datasets if they let you. The point is we don't need huge datacenters in order to train AI models. No massive GPU clusters, no renting compute from someone else's cloud.Just your own hardware. If you have any questions (the technology, the models etc.) I would be happy to answer them.