Izzy Miller

Sessions / 2026 / 2 talks

• Welcome to Our Data Benchmark, Where Everything's Made Up and the Points Don't Matter

  SPIDER, DSBench, and every other analytics benchmark treat data work like a pub quiz: here's a question, here's the answer, did you match? But real analytics is arguing about whether "revenue" means bookings or collections, discovering that Stripe amounts are in cents while your platform stores dollars, and figuring out why the numbers don't tie to last quarter's deck. Current benchmarks can't express any of that—they just check if you got 47.3. Worse: smarter models keep making the same mistakes. Opus is clearly more intelligent than Sonnet, but it falls into the same traps—path of least resistance, accepts the first answer, doesn't ask clarifying questions. We'll show specific examples where industry standard benchmarks fail (including our own) and share some ideas for evals that test what analysts actually do: learn a messy warehouse over time, not answer a frozen question on day zero.

• How Product and Research Build Together at the Frontier

  What do you build when underlying model capabilities keep changing every few weeks? How do you balance long-horizon bets with the pressure to deliver value now? What even is product management in 2026? In this practical, AMA-style conversation, Hex’s AI research and core product leads will share what they’ve learned building agentic products together at the frontier: how product and research challenge each other, make decisions under uncertainty, what has worked (and what hasn’t), and how they turn fast-moving capabilities into complex products that people actually use.