← Collected sources
BUILDERS · EDITED DIGEST

Builders’ Picks | 2026-05-17

2026-05-17 · Historical edition

X / Twitter

swyx, AI Engineer Conference Organizer

swyx previewed AIE, the AI Engineer conference series, coming to India. He shared the news with "AIE coming to India soon," without specific dates or a venue. Given his role at @aidotengineer, this signals the in-person AI engineering conference's expansion into India. For local AI developers, it is a technology event worth watching in advance.

Aaron Levie, Box CEO

Aaron Levie reminded students and universities not to abandon learning and teaching fundamentals across disciplines because of AI. He believes AI can create the mistaken impression that deep expertise is no longer necessary. A knowledgeable professional with AI will always be more effective than a novice with AI. Those who get the most from agents know how to guide them correctly, evaluate output, correct errors, and integrate results into workflows. He gave examples: experienced engineers who have built and scaled complex systems with agents will far outperform people who only vibe code; bankers or analysts who understand financial modeling can accomplish much more with agents. He explicitly rejected Silicon Valley arguments that depth is no longer necessary, concluding that people should keep deepening their expertise.

Garry Tan, Y Combinator CEO

Garry Tan shared two points. First, ZeroEntropy has become his top choice for private AI retrieval across his 120k Markdown personal knowledge base. Second, he sharply challenged Ken Griffin's view of AI, arguing that Griffin has failed to recognize that the ceiling has just risen. He predicted that a person in their twenties reading the post might build an exceptionally capable, symbiotic human–AI team that ultimately surpasses Griffin's entire organization, while Griffin is distracted by cutting costs. His conclusion was direct: "boil the ocean," pursue ambitious goals, rather than simply cutting costs.

Zara Zhang, Independent Builder

Zara Zhang shared a product-building insight: people overestimate how hard it is to make something and underestimate how hard it is to gain attention once it exists. In other words, building itself is often not the biggest bottleneck; winning users' attention and mindshare afterward is. This is a reminder about the tendency to emphasize building over distribution in today's AI startup environment.

Nikunj Kothari, FPV Ventures Partner

Nikunj Kothari warned employees considering newly minted unicorns that they face a triple hit. First, tranched valuations are now common, meaning an employee's entry strike price at the headline valuation is often more than twice the lead investor's preferred-share price. Second, the strike price and 409a valuation will therefore be high: exercising early requires considerable cash, while not exercising early means waiting for the company to grow into that valuation. Third, equity packages look attractive but are heavily inflated by tranched valuations, making nominal total compensation high while actual compensation, at least for now, is not. This is also one incentive for companies to pursue high valuations: raising money with less equity. He advised employees to be vigilant and use Claude or ChatGPT to model their equity's exit outcomes. Joining a unicorn also offers learning, experience, mission, and team value, which should be the core reasons, he added. But employees should conduct investor-like due diligence and not treat all similarly valued companies as equivalent.

Aditya Agarwal, South Park Commons Partner

Aditya Agarwal reflected on meaning in the AI era. On Friday, he wrote a great deal of code using AI; on Saturday, he watched humans play soccer in the FA Cup final and dance in his child's recital. His conclusion: he is not worried that people will lose meaning and purpose. This is an optimistic response to widespread anxiety about AI eroding life's value, emphasizing that human creativity, competition, and emotional connection carry meaning in themselves.