AI Costs Lead Westpac to Prod Staff Toward ‘Sensible’ Model Use
TL;DR
Westpac Banking Corp. is tightening cost controls around internal AI use, according to Bloomberg. The Australian bank is tracking token consumption more closely across the company to understand where expensive model usage is happening. Simpler tasks are being routed toward cheaper models instead of defaulting to the most capable systems. The message to staff is not to stop using AI, but to use it more sensibly: match the model, context size and spend to the job.
Nauti's Take
This is a sober but useful reality check for the AI hype cycle. Productivity does not automatically rise just because employees throw premium models at every small task.
The stronger teams will be the ones that build routing, standards and measurement into everyday workflows. Westpac does not sound especially visionary here; it sounds practical, and that is exactly the point.
Briefingshow
Westpac points to a broader enterprise AI shift: usage growth alone is no longer enough; cost discipline is becoming part of deployment. Tokens are turning into an operating metric, not just a technical billing unit. That matters especially in banking, where scale, compliance and budget control have to work at the same time.
Sources