CBA
Q3CBA's occupied grid is unchanged. The FY26 scorecard defines an 86 percent agentic-messaging resolve rate as conversations closed without a human assisted servicing pathway, and that is the Q3 loop. Q6 stays on the grid as a delivered item, "multi-agent fraud rule optimisation", with no outcome or rule change attributed to it. ChatIT is no longer the named Q5 instrument. Companion is still a pilot.
Trajectory: Q3 deepens. Q6 stays attested until a fraud outcome is attributed to the optimiser.
On 12 August 2026 the FY26 results said about 80 percent of staff actively engaging with AI platforms, including ChatGPT Enterprise or Copilot. No sentence names an internal loop, so none of this lights the grid.
The lines this reading rests on. A date is the source date, when the page or the line names one.
Agentic messaging resolve rate 86% | Improved banker productivity +29% | Increased change velocity +64% | Fewer major incidents -45% | Faster restore time -80%" (FY26 KPI scorecard; footnote 1: "Percentage of customer conversations initiated through the agentic chatbot channel that are successfully resolved without a human assisted servicing pathway.")
Multi-agent code and vulnerability patching" / "Multi-agent fraud rule optimisation to help keep customers safe" (FY26 delivered-items slide)