Your Own Computer Can Now Sort Spam With AI — For Free

A developer has shown that any open-source AI model can double as a lightning-fast yes-or-no decision maker, and the whole thing runs on your own machine. Instead of paying for an AI service or sending your data to a big tech company, you download a free tool called llama.cpp, load any AI model, and simply ask it a question like 'Is this email spam? Answer only 1 for yes or 0 for no.' The AI replies with a single character — but behind the scenes it also quietly reports how confident it was, giving you percentages like 94.9% spam versus 5.1% not spam.

The clever trick is that the AI isn't writing paragraphs or 'thinking' — it's just making one instant guess. On a mid-range gaming computer with a 4GB graphics card, that guess takes about 634 milliseconds, roughly half a second. On a high-end business server chip, it could drop to well under a tenth of a second. For sorting email, flagging bad reviews, or triaging support tickets, that's fast enough to feel instant.

The same approach can handle several questions at once. By encoding answers as combinations of 1s and 0s, you could ask 'Is this spam? Is it a phishing attempt? Is it urgent? Is it dangerous?' all in one go. The trade-off is that this only works for clear, binary-style decisions — it's not the AI writing your emails or answering complicated questions. You're essentially using the model as a very smart sorting machine, not a conversational assistant.

The bigger picture: this puts genuinely useful AI filtering in the hands of anyone with a laptop, no subscriptions, no cloud bills, and no privacy worries. Your inbox, documents, and messages never leave your desk.