Investing in AI
A few weeks ago, I attended UVA Venture Capital Conference, which was in UVA Darden's (their business school) beautiful building in Arlington, overlooking the DC skyline.
The event was great, but the session that really stuck out to me was the Investing in AI session -- the panel was:
Some great points they made during the session:
- As a startup, you don’t need to be sophisticated early — you need something compelling to catch eye or customers, and then you can pivot/expand/mature (As an example Glean had an initial, simple enterprise offering, and how they're pivoting to an agentic AI solution)
- In 2022, there was a crazy explosion of capital, with over 800 new funds being raised
- Venture Capital firms need to offer more than just capital, they need to have platform teams and offerings like SIBR consultant, potential future client introductions, etc.
- The relationship between venture investors and entrepreneurs is so crucial — would you want to have a beer with someone on the other side of your relationship at an airport for a few hours when you’re stuck at the airport? If not, don't establish that partnership.
- They talked about the challenge of AI model poisoning (malicious actors intentioanlly compromising training data that AI models depend on) and model drift (a model becoming less effective over time, such as outside data changing, user behavior changing, or a model overfitting to the data)
- Insight: Junior developers can use GenAI to code, because they're doing common, simpler activities. The challenge for senioer developers leveraging GenAI is that the harder problems aren't so easy for GenAI to recycle from existing codebases (because the problems are less common and more nuanced)
