1) diversify their investments amongst many areas in tech (or maybe some non tech things, not stereotyping all VCS but for this example I'll say diversity of problems being solved using technology/software etc) and thus see commonalities of problems amongst teams in their experience despite whatever the specific niche industry is. I think it is pretty important to agree the cofounders are important. Specifically I've heard from CEOs of startups I use to work for they are really good at hunting out "weaknesses" in the bonds between cofounders.
Basically, they don't want a good idea to fall apart because two cofoudners fundamentally don't see eye to eye, which will make it hard for any decisions they make (whether good or bad) to come to fruition and indicates future struggles with the company, in addition to weak leadership/unified approach when hiring new people, who might be disoriented at understanding the direction the company is going in.
2) Focus on particular niches (a startup who focus specifically on funding AI startups, for example) often they will even fund competitors knowing that if one of them dies, they are almost gauranteed to reep benefits from the remaining on in the niche field. This also means they keep a close track on the variations between two or more companies approach in a specific industry and why one failed and the other didn't.
Someone from Marc Andreessen's team recently published a blog post going into great depth about 5 factors amongst ten or so AI teams that indicated their levels of success in the field, and most of the tilting factors are not what I would say is common knowledge amongst AI startups.
They may have their biases, but they also have unique experience in understanding or seeing repeated dynamics in what could make a relationship (between cofounders) fail (like seeing that one bad relationship where the couple can't see it but it is so obvious to you because youve either now been through it or seen it happen before, imagine this for VCs looking at cofounders over and over again for ten years...) and otherwise have lost lots of money over failed business ideas, so I would say they are incentivized to be as objective as possible with themselves, even if just selfishly for the benefit of their own bank account.
