The modernisation dilemma

Legacy platforms in asset servicing – supporting fund accounting, settlement, reconciliation, and corporate actions – have processed transactions through market crises, regulatory overhauls, and substantial volume growth. Their persistence is not simply inertia. As Thomas Steinborn, Chief Product and Technology Officer at Smartstream, explains, the cost and risk of replacing mission-critical systems have historically outweighed the benefits, particularly where tolerance for disruption approaches zero. When an incumbent system continues to perform the job it was built to do, the case for replacing it is harder to make than it appears.

Operational teams face an additional constraint: the pressure to maintain business-as-usual services for clients leaves little capacity for transformation programmes running in parallel. Where existing technology delivers the required outcome, investment for modernisation must compete against immediate operational demands — and often loses.

Reliability does not mean legacy infrastructure is without cost. Over time, every report, data pipeline, and downstream process built on top of a legacy system creates new dependencies that make change more expensive and more risky. Steinborn notes that manual intervention and break resolution can absorb up to 70 per cent of operational effort in some environments. The institutional knowledge holding these processes together presents its own risk: undocumented expertise accumulated over decades is leaving organisations faster than it can be replaced. “The undocumented expertise that once held these processes together is walking out of the door faster than it can be replaced,” he warns.

Security adds a further dimension that the reliability argument cannot answer. Systems running on operating systems, databases, or middleware no longer supported by vendors can function perfectly well operationally while representing an unacceptable security exposure. A system that works is not necessarily a system that is safe to leave unchanged.

Few firms are choosing wholesale replacement. The dominant approach is incremental modernisation — introducing API layers, cloud capabilities, data platforms, and workflow automation around existing infrastructure without destabilising the transaction-processing core. Steinborn describes this as adding capability without touching the foundations that remain proven and reliable. For exception-heavy reconciliation workflows, the ability to extend rather than replace gives firms a pathway to automation that does not require a high-risk, high-cost migration event.

The industry has learned from large-scale replacement programmes, and the question firms are now asking is not “how do we replace this?” but “how do we make this fit for purpose for where the market is heading?” API-led architectures, hybrid deployment models, and modular software allow capabilities to be adopted incrementally, with each step building on the last rather than depending on a single transformation programme to succeed.

AI is increasingly part of the modernisation conversation, but its role is more limited than the surrounding enthusiasm implies. Steinborn is direct on the point: “AI extends and enhances well-run infrastructure; it does not rescue a broken foundation.” Where data is fragmented and governed inconsistently, AI does not solve the underlying problem — it exposes it. Business logic written years ago by people who have since left the organisation remains opaque: AI can read the code, but it cannot determine whether that logic still reflects current business intent.

Where foundations are sound, the opportunity is more substantial. Smart Reconciliations Air, Smartstream’s AI-native reconciliation platform, addresses exception-heavy processes, accelerates data handling, and supports automated corrections that reduce both operational effort and migration risk. The firms best placed to benefit from AI in post-trade operations are those that have already invested in the data quality and architecture that AI requires to work effectively.

External pressure is making delay harder to sustain. Shorter settlement cycles, real-time processing expectations, and the growth of tokenised assets are placing demands on technology stacks built for overnight batch cycles. T+1 settlement compresses the time available for trade processing, reconciliation, and exception resolution in ways that manual processes cannot absorb. For operational teams that have struggled to build an internal business case for infrastructure renewal, regulatory and market deadlines are providing the justification that internal arguments alone could not.

Steinborn’s conclusion is measured but clear. The market is moving towards immediacy while much of its infrastructure was built for a slower, more manual world. The response is a model of progressive trust – new capabilities earning greater autonomy gradually, with decisions remaining controlled, explainable, and auditable. “In post-trade processing, resilience is the innovation that matters most,” he states. “A system that is faster or richer in features but less predictable is a poor trade.”