How we restructured Airtable's entire org for AI | Howie Liu (co-founder and CEO)

· ai · Source ↗

Summary based on the YouTube transcript and episode description. Prompt input used 79979 of 110717 transcript characters.

Airtable CEO Howie Liu explains how he restructured the company around AI, became the top inference-cost user globally, and why every software CEO must become an IC again.

  • Airtable hit over $100M in free cash flow; the viral ‘Airtable is dead’ tweet used incorrect revenue data from a CB Insights analyst.
  • Liu personally generates more AI inference cost than any Airtable customer globally, running LLM map-reduce jobs worth hundreds of dollars at a time.
  • Airtable reorged EPD into two tracks: a fast-thinking team (ships AI features weekly) and a slow-thinking team (multi-month infra bets like HyperDB, supporting 100M+ row datasets).
  • Liu cut his standing 1-on-1 roster and replaced most recurring meetings with timely, insight-driven sessions to free up IC time.
  • He tells employees they can cancel all meetings for a full week just to experiment with AI products — mandatory play, not optional.
  • PMs must become hybrid PM-prototypers with design sensibility; AEs must become SE-fluent enough to demo without a sales engineer.
  • Liu’s advice to his decade-ago self: never step away from the product details you love, even as the company scales — delegation to execs erodes the integrative thinking that creates non-incremental outcomes.
  • On evals: vibes-based open-ended testing catches things rigid eval suites miss; over-indexing on evals can kill innovation speed.

2025-08-31 · Watch on YouTube