The 2% that consumes 70% of post-trade effort
Around 98% of transactions now clear straight through, matched, confirmed and closed within minutes. The remaining 2% consume roughly 70% of operational effort. These are the mismatches, failed instructions and unexplained cash differences that take days rather than minutes to resolve. In a new article for PostTrade 360°, Thomas Steinborn, Smartstream chief product and technology officer, argues that matching has been automated well while exception management has barely been touched.
Why the exception queue is now unsustainable
T+1 has shortened the window for investigation, volumes are climbing 15 to 20%, and volatility creates mismatches faster than analysts can clear them. ISO 20022 brings richer data and DORA raises the bar on resilience, while hiring and training a reconciliations analyst takes months.
The failure pattern is consistent. Instructions fail while markets keep moving, breaks raised in Asia wait for EMEA to open, and backlogs roll into the next cycle. Each case needs a decision faster than people can make one. Smart Reconciliations addresses the matching side of that problem.
Why AI has not yet closed the execution gap
Some 88% of organisations use AI, yet only around 5.5% report material EBIT impact. MIT’s NANDA research found that 95% of enterprise generative AI pilots delivered no measurable P&L effect. The article’s point is that copilots advise and dashboards report, but nothing acts, so a human still opens the case, queries the systems, decides and closes it.
Exception handling is procedural, documented and rules-bounded, which makes it a natural first home for agentic AI. Smart Agents follows this plan, decide, act, escalate pattern, from detection to closure with humans stepping in at defined risk thresholds.
Autonomy is only worth what its governance is worth
Only 23% of organisations are scaling an agentic system, and Gartner expects 40% of enterprises to demote or decommission autonomous agents by 2027 over governance gaps. Since August 2026, EU AI Act high-risk obligations for logging, traceability and human oversight are legal requirements.
The article sets out three principles: guardrails that keep agents inside workflows and approval limits owned by operations, an immutable audit trail on every step, and graduated autonomy that is revoked if behaviour drifts. Firms taking this route report 50 to 70% reductions in exception-handling effort, with a named human owner remaining accountable.
Read the full article
Read Thomas Steinborn’s full argument in Kill the exception queue: post trade’s real bottleneck is the 2% on PostTrade 360°. To see how agentic AI can take on exception handling in your operations, explore Smartstream Smart Agents.
