Lan Xuezhao Profile picture
PhD in human psychology investing in Machine Intelligence. Founder @BasisSet. Prev. @Dropbox, McKinsey. Building an institution & raising 2 kids.
Nov 17, 2023 13 tweets 4 min read
Can LLMs predict the future for startups? Sounds impossible, but is it?

We've experimented with using LLMs to try to predict company outcomes, based on founder psychological attributes alone. Wow, it worked much better than anticipated! Read more and see your own results. 👇 Image We built a dataset with longitudinal data from 37 portfolio companies. We recorded how they performed over the years on a set of financial metrics. We enriched the data with founder attributes across ten dimensions, mostly taken from our initial exploration of founder psychology. Image
Mar 15, 2022 8 tweets 3 min read
I've seen LOTS of pivots in my very short VC career, but @PeterLZhou and @cyrieu at @RutterAPI took it to a whole new level. I have not seen such consistent monthly investor updates when something does NOT work, for YEARS. 👇 For years, their investor updates were basically "tried A, Didn't work; here's what we learned. No burn. Trying B next".
They are -
1) brutally honest
2) surprisingly consistent
3) after a while...very entertaining to read (I'm sure Peter and Eric felt differently)
Dec 27, 2021 24 tweets 5 min read
One of the most common questions we got from our @BasisSet portfolio companies is how to hire the first VP Engineering: When? What profile of candidates? How do they interview for this role? @niniane is the best CTO I’ve worked with and has seen it all, she wrote about it here 👇 Read her full article here: bsv.medium.com/hiring-a-vp-en…
Mar 25, 2020 14 tweets 6 min read
I posted about sourcing #N95masks for #coronavirus in an earlier tweet, now here’s more about sourcing #testingkits, #ventilators etc. and how I got my kits. Supply-chain is fascinating. Storm 👇 There are multiple approaches to testing. The fastest is an antibody blood test I got. A more accurate approach is RT-PCR via a nasal swab. It’s slower since it requires a lab to process the samples. Many countries use both tests for each patient.
Dec 17, 2019 14 tweets 3 min read
Against Conventional Wisdom, whether a founder/CEO is technical does not differentiate company success. Also, founders of all ages can be successful. However, having a complementary cofounder, often a technical one, is correlated with success. The best founders know their strengths and weaknesses and recruit a complementary team that maximizes the company’s chance of success.