It's absolutely automated. Whole companies specialize in it (Cyveillance is one - look them up) and routinely patrol the web looking for any use of trademarks and then automatically sending out threats.

I worked on a site selling second-hand goods and there were around a dozen robots patrolling the site every day, completely ignoring the robots.txt (but fetching it in order to specifically look at disallowed links first). They homed in on any trademarks and kept fetching the pages repeatedly.

They became such a bandwidth hog that I had to block their IPs (whole CIDR blocks) - if you blocked the user agent, they immediately re-fetch with a faked IE or Mozilla UA and then used that to trawl the site.

Once they have their 'evidence' of the mark use, they would post (regular snail mail) automated threats with the offending web pages printed and the TM terms highlighted. Typically demanded money (but not always) plus a promise never to use the TM ever again and destroy all computer records containing the TM.

Thankfully the shop owners just ignored them as, like the article points out, it's perfectly legitimate to use a trademark in context. Besides in the country concerned it was a trading description offence to incorrectly describe the goods (fines and loss of license to trade in second hand goods). (i.e. if it is a genuine Gucci handbag, you describe it as that but if it is a fake, you must not mention Gucci. Enforcement officers checked regularly that the rules were followed - surprisingly logical and sensible and trumps any nonsense from an automated TM enforcement bot).