Why rent expensive cloud computers by the token when your unused laptop, phone, or browser tab could help power the next AI run? WebAI at Home connects spare devices into a shared network that runs large models through community-owned hardware instead of sending every request to OpenAI or another cloud provider. It is not just another chatbot subscription: it is a open alternative for batch AI work, turning idle devices into useful computing power.
Where it came from
There are two ways to run a large language model today. You can pay a company. You send your text to somebody else's data centre, you accept their terms, their prices, their availability, and their decision about what the model will and will not answer. Or you can buy a graphics card. That works, and it is genuinely yours, but it costs more than most people will spend on a hobby, and it puts a hardware purchase between a person and the ability to compute. Both of these are forms of permission. In the first case you ask a company. In the second you ask your bank. Neither is a technical limit — it is not that the computing capacity does not exist. It is that the capacity which exists is not arranged in a way that ordinary people can use. webai-at-home is my attempt to arrange it differently. It asks whether a group of ordinary people, using the devices they already own, can run a language model together without asking anyone for permission to do it.