Two investors discuss why AI tools struggle at pre-seed stage: startups lack traction to train on, and custom models deliver only about 25% of the value of direct founder conversations. They explore how their 400+ portfolio companies could benefit from internal support instead.
Pre seed.
Also, kind of embedding AI tools is difficult because there's typically no traction whatsoever.
So ingesting.
Any type of metrics or traction into a model that you've customized is. I don't, I don't know what the output. I mean, I played around with it, but ultimately it lands out in stuff that's probably like 25% of the way that I could get it if I just chatted with the founder directly.
It does open the door for opportunities to support founders within our portfolio, and I think specifically within our own team at 400 plus companies. And I know we've talked about, like, what are some ways that maybe we can bubble up insights about perception of investors when our founders want to go out to raise.
