Our IT manager is constantly pitching me on building expensive custom machine learning models, but I suspect it is just software theater. How do we structure these initiatives so they are treated as practical operations-improvement projects rather than high-risk technology experiments?

Question

Tyler Smith · Accepted Answer

To avoid expensive science experiments and software theater, you must change how you frame these initiatives. Never pitch or fund an AI project. Instead, pitch and fund operations-improvement projects that happen to use machine learning as a backend tool.

When your IT manager or software developers pitch a new technology, force them to translate their technical jargon into clear business metrics. Do not let them talk about neural networks or large language models. Instead, ask them which specific seat on your Accountability Chart will benefit, which core process will be streamlined, and how many hours of manual labor will be eliminated.

Every technology initiative should be run as an operational Rock. It must have a clear owner, a defined scope, and a measurable outcome that directly impacts your weekly Scorecard. If the project cannot be tied directly to reducing costs, increasing capacity, or improving service delivery quality, do not fund it.

By treating AI as a footnote rather than the headline, you keep your leadership team focused on what actually matters: building a highly efficient, system-dependent business. You want practical solutions that solve real operational bottlenecks, not flashy tech demos that drain your cash reserves.