Yeah I agree, maybe I should have said if it's not in your "Top 3"
I think it also depends on your resources. If you're a solo founder then it can be hard to justify much time on social media.
And I also think you can have both. You can automate some posts, and also do ad hoc stuff. In my use case I publish articles on a consistent basis, why not automate the social media side of that?
I got >100k genuine users in my facebook group, 10k users on instagram etc and now I have stopped automation on both as the genuine user interaction suffices for growth.
Got API privileges revoked by twitter and the moral of the story if you have to be a 'crook', go all in ;)
ps: my first version of the project is here: https://github.com/ganeshkrishnan1/socialmediabot
Now I use scala, tensorflow and stanford NLP
This is the first point under the "There are some downsides to automating" section. I think that's the most important thing to read and realize whether you should even continue to read or not.
If social isn't a marketing strategy your startup is using at all, yeah, do this. It doesn't really matter too much then. But if you're actually trying to use social for any value, doing this is a terrible idea, in my opinion. The author/OP phrases it as this is all a good idea if social isn't your main marketing/acquisition channel, but I think this is still a bad idea if social is even your secondary or tertiary marketing/acquisition channel. Your engagement will tank on social, and people will be able to smell it a mile away if your social is worth engaging with or not. It's quite easy to tell if stuff like this is automated or not. I feel like the OP downplays that too much.
With that said, you do you, and decide what works best for your company. ¯\_(ツ)_/¯ This is still a useful post if you put what I mentioned above aside, and decide to do this for your social strategy.