By Gaurav Mehta, CCO, NOVA CMX
Ask any operations manager at an Australian broking firm about their afternoon routine and you’ll get variations on the same story. Exception queue. Settlement cut-off. A spreadsheet that has somehow become a mission-critical system. And a team making rapid judgement calls about which breaks matter, which can wait, and which are quietly becoming tomorrow’s problem.
That’s the operations floor AI is beginning to change. Not through some wholesale replacement of experience and judgement, the industry has learned to be sceptical of that promise, but through a more targeted question: which of these decisions actually requires a person?
Four areas in post-trade are seeing genuinely practical AI application. All four share a common principle: AI earns its keep on volume, pattern recognition and speed. Humans earn their keep on context, relationship and judgement.
Exception triage
Post-trade operations teams don’t just manage exceptions. They manage the decision about which exceptions to manage, and that decision, repeated dozens or hundreds of times a day, consumes significant time and expertise. AI changes the nature of this work by classifying exceptions before a human touches them: probability of self-resolution, dollar value at risk, settlement deadline proximity, counterparty history. The result is a prioritised queue rather than an undifferentiated stack. Operations staff direct their attention to exceptions that genuinely need it, rather than working through low-risk items first simply because they appeared first.
The caveat is real: AI triage is only as good as the training data behind it. Novel exception types, the ones that haven’t appeared before, still require experienced human assessment. The tool sorts the queue; it doesn’t replace the person who resolves the hard ones.
Reconciliation
Traditional reconciliation is rules-based: exact match first, then manual break resolution. AI augments this by identifying patterns in break history, which counterparties generate which types of breaks, which instruments are consistently problematic at month-end, which upstream data quality issues drive recurring mismatches. This shifts reconciliation from reactive break resolution toward earlier identification of systematic problems. The operational value is not only faster resolution; it is finding the root cause before it creates next week’s breaks.
In an Australian market operating across ASX and Cboe Australia, where multi-venue position management is a daily reality, this pattern recognition is increasingly valuable. Transaction volumes in post-trade operations have been growing materially year-on-year across the industry, and reconciliation workloads are following.
Corporate actions processing
Corporate actions remain one of the most manual, deadline-sensitive workflows in post-trade. Rights issues, schemes, special dividends and off-market transfers require accurate instruction capture, entitlement calculation and client notification, often within tight market deadlines. AI can monitor event calendars, validate incoming data against expected parameters and flag discrepancies before they become errors. The European Central Bank has identified processing errors as a primary driver of settlement failures in securities markets, and corporate actions events are a material contributor to that category.
Where AI should not fly solo: complex events with multiple client elections, unusual event structures or decisions with a client relationship dimension. These remain squarely in human territory.
Settlement fail prediction
This is where the commercial case for AI is most direct. A settlement fail that has already happened costs money, in buy-in exposure, operational remediation and counterparty friction. A fail predicted twelve to twenty-four hours ahead can be interrupted: securities sourced, funding arranged, counterparty contacted. AI models trained on historical fail patterns, by counterparty, security type, settlement period and market conditions, can flag at-risk positions while there is still time to act. For Australian brokers managing multi-market settlement, that early warning function has real economic value.
Where human judgement still runs the show
AI in post-trade is genuinely useful. It is not yet genuinely autonomous. Credit decisions, client-facing break resolution, novel event types, regulatory judgement calls and anything requiring counterparty relationship management remain the domain of experienced operations professionals. The honest design principle is that AI handles the volume and the pattern; the human handles the consequence.
For Australian operations teams facing rising transaction complexity and constrained specialist headcount, the practical value of AI is not replacing people. It is making the people you have significantly more effective and letting them spend their time on the work that actually requires them.
The spreadsheet-as-mission-critical-system has had a good run. Its successor should probably know the difference between a break that matters and one that doesn’t.
NOVA CMX provides post-trade automation and operations technology to sell-side financial institutions across Australia-New Zealand, Asia and the UK.
