Notes: By it's very nature, auto-spintaxing produces cases of nonsense grammar. The English language is incredibly complex, and uses words for far too many meanings, especially with the colloquialisms that have become common in everyday language today. It's impossible (without full language processing) to cover all phrasing scenarios, and A.I. language processing is an enormous computational task - certainly well beyond what the typical web-server or PC could handle in a sensible time-frame.
Our software tries to pick out certain common phrases - as well as individual words, but inevitably you'll end up with cases where a word is used in an alternate meaning to the 'norm'. We do synonym database checks to ensure that phrase replacement works in 75%+ of normal language cases, but this still leaves plenty of room for error. If you're just producing articles, then with the amount of non-english speaking writers working today, it probably won't matter too much. But if you want it to read well, (and maybe pass a human audit,) then you'll normally need to whip through and tidy-up some of the spin outputs.
Also, bear in mind that technical or industry-specific terms become problematic - as they're often used in alternate phrasing formats.