Just Software: An Advisor's Case Against the AI Hype Cycle

A multi-client AI advisor sees most enterprise AI as hype measured by the wrong metrics, with real value confined to multi-parameter use cases and progress blocked by immature governance and decision paralysis.

One independent advisor working across consumer products, manufacturing, and technology firms offers a deliberately deflationary view of enterprise AI. He estimates roughly 80% of cases are still noise and begins client work by reframing AI as 'just software' with defined limits. The recurring failure is measurement: firms apply old KPIs to new tools, so recruiters scan hundreds of applicants while time to hire barely moves. Throughput rises; value does not. He sees durable payoff only where AI handles many parameters at once, such as per-client product design replacing generic personas, and warns that one-shot automations decay and become maintenance burdens. Compute spend is a warning light: software teams exhausting a year's token budget in a month signal an exhausted operating model, prompting experiments that merge engineering with product. Governance scores a 2 to 3 of 5. Policies exist but maturity does not, and restrictive lockdowns that limit staff to basic tools are immaturity dressed as control, pushing real building to outside consultants. His sharpest point: AI hype is causing corporate decision paralysis, with leaders deferring choices and businesses quietly suffering, while most actual layoffs are not AI-driven. He wants someone to publish a plain-language definition of what an agent is, to stop anthropomorphism from inflating expectations. The throughline is literacy before deployment, outcomes before activity, and enabling governance before restriction.