I am pretty interested in what the person letting it run 24/7 is achieving. Is it a continuously processing workload of some kind that pipes into the model? Maybe a 24/7 chat support with high throughput? Very curious.