The Bank Still Owns the Mistake
Westpac shows the real banking AI test: faster work without compounding risk.
Westpac’s GenAI story is not a chatbot story; it is AI moving into the regulated work behind the banker.
The COO question is whether AI has changed cycle time, rework, handoffs and customer outcomes.
The CRO question is sharper: when AI drafts, summarises or recommends, who owns the mistake?
Most people imagine banking AI as a chatbot.
That is the wrong place to look.
The more interesting Westpac story is happening behind the banker: AI drafting, summarising, searching, triaging, surfacing information and supporting lending workflows before a customer ever sees the output.
Westpac has rolled out Microsoft 365 Copilot to 35,000 employees, contractors and service providers after a 15,000-person pilot. It has also been reported to be using Copilot Studio for internal GenAI agent development, while iTnews has reported that Westpac is looking at AI across business banking workflows including documentation, KYC, annual reviews, credit memos and customer communications.
That is real activity.
But it is not yet proof of transformation.
It is proof that AI is moving into the seams of banking work.
Before you read on, where is AI already drafting, summarising, recommending or prioritising inside your organisation before anyone senior sees the output?
That is where the real governance question starts.
Below, I unpack why Westpac is a useful COO/CRO case study, and why “the bank still owns the mistake” may be the most important AI principle in regulated industries.
