Long-form interviews and conversations with tech, AI, and finance leaders (30min+)
Robert Brennan outlines the evolution from single AI coding agents to parallel agent orchestration, demonstrating how breaking massive refactoring tasks into coordinated sub-tasks can deliver 30x productivity improvements on tech debt remediation while maintaining essential human oversight.
Michael Bloomberg shares his journey from a laid-off Salomon Brothers partner to building a financial data empire, emphasizing that success comes from solving real problems, outworking competitors, and taking care of people—principles he applied equally to business, 12 years as NYC Mayor, and giving away over $20 billion in philanthropy.
SallyAnn DeLucia and Fuad Ali from Arize demonstrate how iterative "prompt learning"—combining automated evaluations with human explanatory feedback—can improve AI agent performance by 15% without fine-tuning, outperforming traditional optimization methods while reducing costs significantly.
Peter Wielander demonstrates how Vercel's open-source Workflow DevKit transforms local AI agents into production-grade systems by wrapping AI SDK code in durable, retryable workflows with minimal refactoring, enabling automatic observability and resumable streams without managing queues or databases.
![Claude Agent SDK [Full Workshop] — Thariq Shihipar, Anthropic](https://i.ytimg.com/vi/TqC1qOfiVcQ/hqdefault.jpg)
Anthropic's Thariq Shihipar introduces the Claude Agent SDK, which packages the Claude Code architecture to build autonomous agents using a 'bash is all you need' philosophy, leveraging Unix primitives and dynamic code generation instead of rigid predefined workflows.
Ryan Kidd of MATS discusses high uncertainty around AGI timelines (median 2033), the paradoxical state of AI safety where frontier models display both surprising ethical alignment and concerning deception capabilities, and why MATS employs a portfolio strategy to develop talent across diverse research agendas—including long-term bets that could be compressed by future AI labor.
Whole Foods co-founder John Mackey shares how missionary-level dedication, avoiding venture capital control, and capitalizing on competitors' 20-year distraction allowed him to scale from a single hippie grocery store into a national chain that changed American eating habits.
Exor NV, the Agnelli family’s century-old holding company, trades at a 60% discount to its net asset value, effectively allowing investors to buy shares of Ferrari—its crown jewel holding—at roughly half price while receiving stakes in Stellantis, Philips, and luxury brands essentially for free, though the eclectic portfolio mix and capital allocation concerns create legitimate risks.
Luke Drago introduces the 'intelligence curse'—an economic phenomenon analogous to the resource curse—where reliance on AI as the primary factor of production eliminates incentives to invest in human capital, potentially destroying the economic bargaining power that underpins democratic rights and individual agency while concentrating dangerous levels of power among those who control the technology.
Mathematician Joel David Hamkins explores how Georg Cantor revolutionized mathematics by proving that infinities come in different sizes, from the counterintuitive countable infinity illustrated by Hilbert's Hotel to the strictly larger uncountable infinity of real numbers revealed through the diagonal argument.
![[State of Research Funding] Beyond NSF, Slingshots, Open Frontiers — Andy Konwinski, Laude Institute](https://i.ytimg.com/vi/ZagdY6UJYL4/hqdefault.jpg)
Andy Konwinski details the Laude Institute's dual-track model to revolutionize research funding by combining a nonprofit for open-source AI with a venture arm for PhD-led startups, aiming to replicate the 'Databricks Motion' where breakthrough research becomes billion-dollar companies through high-velocity, Silicon Valley-style capital deployment that complements underfunded NSF programs.
![[State of MechInterp] SAEs in Production, Circuit Tracing, AI4Science, "Pragmatic" Interp — Goodfire](https://i.ytimg.com/vi/CoZp6xaTbWw/hqdefault.jpg)
Goodfire researchers discuss how mechanistic interpretability has evolved from pure research to practical deployment in 2025, highlighting production applications like PII detection and scientific discovery while navigating the field's pivot toward 'pragmatic' tools that prioritize real-world utility over complete mechanistic understanding.