AI-generated memos kill the filtering power of Amazon's PRFAQ ritual, Vets say
brandon_galang · x · 2026-10-26
Jean-Denis Greze (ex-Amazon) argues that forcing 6-page memos was a filter: bad ideas got exposed by the painful writing process itself. Now AI can produce polished memos without real thinking, so the filter is gone — and in a company like Amazon where people are incentivized to get initiatives approved, some will inevitably game the process with AI.
PM Brandon Galang agrees: he came up through PRFAQ, still uses the working-backwards principle for building agents, but as a corporate ritual it no longer serves its original purpose.
More from AGI Musings
- Immunologist Derya Unutmaz: AI progress means PhD programs should now be cut by at least 80% — TinfoilTricorn · 2026-10-26
- Gary Marcus mocks betting the economy on an industry 'beset with thieves and grifters' — GaryMarcus · 2026-10-26
- Token usage doubling every 25 weeks could take AI from 0.1% to 4% of GDP in ~2 years — robleclerc · 2026-10-26
- ML papers look intimidating because authors write to pass review, not to teach — TivadarDanka · 2026-10-26
- VraserX: the century's biggest medical breakthrough may be continuous body repair — VraserX · 2026-10-26
- Sam Altman: robots building robots and data centers building data centers form another AI flywheel — haider1 · 2026-10-26