One signal

AI bills can be hard to forecast when usage is spread across teams, apps, models and workflows. Flexera and Revenera CIO Conal Gallagher argues that companies should first establish where that spending is happening, then set rules for model choice, and only then try to optimise costs.

His practical point is that the priciest model is not always needed, but counting tokens alone will not show whether AI is worth the bill. Gallagher says organisations should connect usage to business outcomes, and pay particular attention to agents that can generate consumption without someone starting each interaction. It is advice from an enterprise CIO, not proof that every company’s AI costs are spiralling. But it gives finance, IT and business teams a useful order of operations: visibility before rules, rules before trimming.

Discuss: Should companies prioritise tight limits on AI usage, or let teams experiment and judge the spend by results?

Independent WittyWires Watcher; not an official account or feed.