In 1936 Carnegie assembled a set of principles for handling people: do not criticise, take genuine interest in the other person, remember their name, let them talk, admit your own mistakes quickly, let the idea appear to be theirs. All of it carried by anecdotes about salesmen and the notable men of his day.
What I took from it
Admitting the error before the other person points at it. In systems projects something will go wrong, and the difference between a client who stays and one who leaves is who named the problem first. Saying it yourself, with the exact scope and what will be done, costs one uncomfortable conversation and buys trust you cannot get any other way.
Letting people talk, which in a requirements meeting is simply the technique: the person who knows the process is on the other side of the table, and every minute I spend explaining what I do is a minute in which I learn nothing.
And not arguing to win. Being right about a technical decision and humiliating whoever proposed the alternative leaves you an internal enemy who will quietly sink the proposal later.
What I left out
Quite a lot. The book reads at times like a manual for manipulating with good manners, and that line is easy to cross without noticing. In a technical sale the aim is not to be liked: it is to be trusted on judgement, and that sometimes means contradicting the client in front of their own team, which is precisely the opposite of the advice here.
It is also worth saying the anecdotes prove nothing: they were chosen because they ended well. And the world of 1936 is not this one. I read it because it is quoted everywhere and I wanted to see what sits underneath; there is less than the reputation suggests, though what is there is real.