Domo CDO argues AI adoption driven by fear and impatience produces theater, not results, and companies should start with small, scoped use cases.
Key Takeaways
LLMs have no product spec: “it’ll do anything for anyone” creates strategic confusion for organizations trying to deploy them.
“Tokenmaxxing” – buying model access and maximizing usage – inflates activity metrics without moving business outcomes.
Klarna’s cycle of replacing customer service staff with AI then rehiring humans is cited as a cautionary example.
Willis recommends starting with narrow, verifiable automation (e.g., invoice anomaly detection) where human judgment handoffs are explicit.
CFOs are beginning to challenge AI spend with no measurable return, signaling a budget reckoning ahead.
Hacker News Comment Review
Commenters broadly agree with the fear-driven FOMO framing; veterans of pre-LLM deep learning note the hype pattern is familiar and not new.
Outside the SF/finance bubble, commenters observe a detectable mood shift over the last six months: excitement has curdled into annoyance at AI slop and half-baked deployments.
Notable Comments
@cmiles8: notes the bubble appears “largely ignorant” of the broader mood shift already underway beyond Silicon Valley.
