How it works
A marketplace that turns idle computers into capacity for AI work.
Some AI work does not need an answer this second: transcribing recordings, extracting information from documents, preparing answers for later. This is called batch inference, and it is a good match for computers that are only busy part of the time.
Submit
A business sends a batch of work through a web portal or an API. HCX confirms what the work involves before anything starts.
Distribute
Our software splits the batch into many small, independent jobs and schedules them across enrolled computers, wherever there is spare capacity.
Run
Enrolled gaming PCs run the jobs when their owners are not using them. The owner always has priority; work yields the moment the PC is needed.
Check
Every result is checked before it is delivered, and carries a record of how it was produced.
Deliver
Results come back through the portal or API, and the customer is billed for work that was delivered.
Built for lighter impact
Using computers that are already built, already connected and often idle, means less new hardware and less new demand on the electricity grid. That is the idea behind “kind to the planet”.
What we do not do
HCX is for batch work, not live chat. We do not train models. We do not claim to protect confidentiality against the owner of the computer a job runs on, so customers should say what kind of data they submit and we will tell them how it can be handled.
Looking for detail? See Customers or Documentation.