Not Every Problem Needs an AI Agent
CIO, Tuesday, August 18th, 2026
A former ML platform leader argues the smartest AI system is often the one that knows when not to use an agent.
The author led a large team of data and machine learning engineers that had built a full ML platform from scratch with dozens of models in production delivering measurable results.
When generative AI arrived, board pressure changed the question from whether the technology could help the product to how fast a place could be found for it.
The article describes what happened to a working recommender system under that pressure. It argues the smartest AI system may be the one that knows when not to use an agent at all. The piece is a caution against retrofitting agents onto problems that classical ML already solved.