AI Costs Lead Westpac to Prod Staff Toward ‘Sensible’ Model Use
TL;DR
Bloomberg reports that Westpac Banking Corp. is tightening control over its AI costs. The Australian bank is tracking token usage across employees and systems company-wide. Simpler tasks are being pushed toward cheaper models instead of sending every prompt to expensive frontier systems by default. Management is framing this as 'sensible' model use: AI adoption continues, but model choice is increasingly sorted by effort, risk and business value.
Nauti's Take
This is the boring but important phase after the AI rush. Companies are realizing that 'everyone uses the best model' is not a strategy, it is an expensive default setting.
Westpac is turning that into management discipline: route simple tasks cheaply and reserve premium models for cases where accuracy, context or risk justify the price. That is what mature workplace AI starts to look like.
Briefingshow
Westpac shows where enterprise AI is heading: not toward less usage, but toward cost control per request. Tokens are becoming an internal operating expense, much like cloud compute. Teams that scale AI now need model routing, usage telemetry and clear rules for which work actually deserves an expensive model.
Sources