Ask anyone who runs servers what happens when one dies in the middle of a job. The job dies with it, something has to notice, and the work starts over from the beginning. That's the rule this 42-second demo breaks, on camera.
Four servers are working together on a single AI answer. Partway through, we shut one of them down outright, and you can see the command being typed and executed. The remaining servers see the drop and keep working. The answer finishes anyway, and the status panel on screen shows the system healthy the whole way through.
Nothing restarts and nobody steps in. TAHO runs beneath your existing infrastructure and handles a lost machine as a normal event instead of a disaster
