Ollama
2.5 hours of busywork, handed off free
A freelancer who sells automation services on Fiverr was spending almost three hours a day on mechanical checks: three email accounts, his Fiverr inbox, calendar, weather and code reviews. He built a set of Python scripts on his always-on Mac that use a local Qwen model through Ollama for the judgment calls and plain cron for the rest, sending him Telegram alerts only when something matters. The system never replies, books or posts; it surfaces, he decides.
How it wentAfter three months he reports about 2.5 hours a day saved, roughly 75 hours a month, at $0 for the AI layer, with about 12 hours of setup and 30 minutes a week of maintenance. The Fiverr filter caught 100% of scam messages with no false positives in his manual checks, saving about 10 hours a month on that inbox alone. It broke along the way: expired Fiverr sessions, alert fatigue, and hallucinated calendar events, each fixed with a narrower rule or a verification step.
The prompt · from the source
Is this a legitimate business inquiry or a scam?
Scam indicators: external contact requests, generic greeting + immediate project offer,
poor grammar with urgent tone, requests to communicate off-platform.
Message: {message_text}
Respond: LEGITIMATE or SCAM with one-sentence reasoning.