Regular expressions in keywords matching?

Worth knowing before you wait for it: regex wouldn't actually solve the typo half of this. Regex matches patterns you define in advance, and typos aren't patterns, they're random. You'd be back to enumerating variants, just with more syntax. Edit distance matching is what handles typos, and regex isn't that.

What does fit both of your problems is intent matching. Instead of listing keywords you write a description of what the person means, and ManyChat's AI matches against it. That covers typos without you predicting them, and it handles your Messenger case too, since "thank you" with six emoji and "thanks so much!!" mean the same thing and an intent catches both where a keyword list never will.

It also solves the part you're most worried about. Haters and people leaving a side note don't match the intent, so they don't trigger and don't burn your limits.

Test it on one automation before moving everything across. Intent matching is less predictable than a keyword list by nature, so you want to watch how it behaves on your real comments first.

Thank you for your hint, but I am sure if I have a word “zánět” and people type “zánet, zanět, zanet, zánét etc.” I can easily match this by just doing z?n?t. It’s this type of ‘typos’.

If Manychat’s AI is as clever as Meta’s AI, then thanks but no thanks :)

Hi ​@NicoleZvR, as far as I know regex isn't supported in keyword rules right now. I'd post it in Ideas so the product team sees it and others can vote for it:
https://community.manychat.com/ideas?sort=votes.desc

If you need it sooner, you could run the regex outside ManyChat with an External Request and use a Condition to decide if the reply goes out.