It makes a lot of sense. Imagine you have a fund that you need to invest into new, unproven companies at the early seed stage. You can find companies the tradit

It makes a lot of sense. Imagine you have a fund that you need to invest into new, unproven companies at the early seed stage. You can find companies the traditional way by having meetings, getting referrals, soliciting applications, etc. By the end, you'll have had a total of a couple hours (at most) contact with the founders before you decide to invest. On the other hand, you can start out by spending 200k on 20 companies for the summer while mentoring them. By the end of the funding cycle, you'd have a really good idea of which founders were hustling, which ones suck, and which ideas you believe in. Based on this unusually in-depth knowledge of the companies and founders, you can invest more in the ones you like. Presumably, this would improve the quality of your overall portfolio.