Members of SV Icons are huge fans of Elad Gil, and it was a great honor to host him for dinner last week. We so appreciate his time and willingness to share insights with us, which I have summarized below. Thank you so much, Elad!
Biggest Untapped Applications for Generative AI:
· Consumer Social (like Can of Soup)
· Copilots for white-collar services (like Harvey)
AI Apps Will Create Defensibility in 3 Ways:
· Owning a workflow, fine-tuning their model, and acquiring special user data
How to Identify the Next Big Trends:
· Chase technical interestingness and innovation, not money
· Develop a diverse network of smart people
5 Use Cases of Crypto/Blockchain:
· Generational replacement for gold
· DeFi (wire money 24/7)
· Privacy
· Art
· Digital identity (cryptographic proofs to authorize AI agents to share info on our behalf)
What We Underestimate:
· We often underestimate TAM. Building a product with high ease of use can greatly expand the market (Uber for example).
· We also underestimate the massive resources put in to optimize transformers for GPUs, which leads to a powerful hardware and software lock-in effect.
· Transformer architecture is likely not even the best: Many models are sitting out there that no one is testing/using because they too expensive to train.
Future of AI:
· Everyone (consumers, corporations, governments) will have multiple AI agents.
· We are still likely 2+ years out from real LLM adoption by enterprises given long planning, testing, and building cycles. Usage and hype will only increase.
How Elad Invests:
· While most investors focus on the team at the early stage, Elad focuses on the market.
· Cited Andy Rachleff: “The #1 company-killer is lack of market. When a great team meets a lousy market, market wins. When a lousy team meets a great market, market wins. When a great team meets a great market, something special happens.”
Advice for Founders:
· Be default building. Even at the earliest stages, the best founders are always making prototypes, iterating, and experimenting.
· The best founding teams always have someone who is great at building and someone who is exceptional at selling.
· Sometimes, the best marketing, not the best product, wins.
· Good people extrapolate their behavior onto bad people. Don’t do that. If someone gives you a reason to mistrust them, do so.
· Your GTM strategy should be a function of your business model (pricing, gross margins, average contract value). Many businesses cannot support sales reps.
· Need to put yourself into mindset of person sitting across from you in a business context. Figure out what their incentives are and give them what they want.