Executive Voices“The change I most want to see is the end of reconciliation as an industry” – Steven Croker

Much of post-trade still comes down to reconciliation: do two records of the same trade agree? Across the industry, thousands of people are employed to answer it.

That model is under strain. T+1 has compressed settlement cycles, AI is changing what operations teams need to look like, and tokenisation promises to rewire collateral and custody. Yet many banks still run on systems built decades ago and on handoffs between front, middle and back office.

Steven Croker, an IB & Markets Operations Leader, has spent more than a decade inside that problem, most recently leading a team of around 220 across the GCC, Europe, Asia and North America. In this conversation with Digital Trade Outlook, he talks about what has genuinely been automated, how ready the region is for accelerated settlement, where DLT earns its place, and why he wants to see the end of reconciliation.

Digital Trade Outlook-You have spent your career in the engine room of banking: middle and back office operations across FX, derivatives, structured products and securities services. How has the perception of operations changed over that time, from cost centre to strategic function?

Steven Croker, IB & Markets Operations Leader: Operations has evolved from cost centre to strategic enabler. The shift is real, though it’s perhaps taken longer than it should. The transition happens when organisations start measuring operational resilience alongside market performance. What’s changed is the framing; what’s still emerging is the investment model: most institutions haven’t fully internalised that operational capability requires the same talent and capital allocation as front office.

The industry has talked about post-trade automation for two decades. What has genuinely been automated, what remains stubbornly manual, and where do banks most often underinvest?

As an industry, we have genuinely automated the deterministic workflows: matching, high-volume reconciliation, regulatory reporting. The stubborn manual segments are exception handling and complex collateral and more structured cash flow movements, not because automation is impossible, but because they sit at the intersection of multiple systems with different design philosophies, plus, relevant to the region, a buy-side reluctant to utilize industry tools that broker-dealers continue to push. The real opportunity, though, isn’t incremental tool deployment; it’s rethinking operations from an end-to-end lens at the design phase. Too many organisations automate in silos, optimising front-end UX without architecting how data flows through settlement, collateral, and custody. That integrated design approach is where most banks underinvest.

T+1 settlement in the US has reset expectations globally. How prepared are Middle East institutions for accelerated settlement cycles, and where does the region’s operational capability stand against hubs like London and Singapore?

T+1’s real operational challenge emerges during volatility and lower trade flows, when collateral efficiency and exception handling can’t be deferred. Middle East banking institutions have strong technical foundations, but the operating model maturity required for compressed cycles under stress remains the gap. It’s a 9–12 month operating model refresh focused on collateral orchestration and peak-load workflows, not a settlement parameter change. The region has the capability to close that gap faster than others, but it requires treating it as a redesign exercise.

Where do you see the real use case for DLT and tokenisation in post-trade: settlement, collateral, custody? And where is it hype?

The most viable use case is collateral optimisation: intra-day reuse, cross-border movement efficiency, and reduced settlement friction. Settlement on DLT has merit, but it depends on ecosystem coordination. Custody remains early-stage. The honest view: DLT solves real problems in specific workflows, but only if those workflows are redesigned end-to-end to leverage the technology. Point solutions often fail because they’re grafted onto legacy operating models.

Middle and back office talent is a persistent challenge: the work is complex, the profile is low. How do you build strong operations teams, and how is AI changing what those teams need to look like?

This is where the real tension sits in the evolving workplace. The talent profile has already shifted from task execution toward systems design and judgment. As AI handles repetitive work, operations teams need architects who understand interconnected processes, not just individual steps. But here’s the constraint: many data and AI analytics departments are drowning in legacy modernization projects. Legacy systems created between the 1970s and 1990s still power many global banks, built for stability but not designed for real-time analytics or intelligent decision-making, forcing AI teams to spend cycles on data quality, integration, and infrastructure rather than building differentiated models.

The economics matter too. Self-service operational analytics platforms are emerging as the solution: empower middle and back office teams to configure and run AI models without waiting for data science bottlenecks. But compute usage and model ROI require discipline. Anthropic has deepened its push into Wall Street, focusing on infrastructure and deployment mechanisms for banks to configure and run AI agents themselves, and with that comes the discipline of ROI tracking, not just model performance. According to a recent Unframe report, most institutions still struggle to convert AI maturity into measurable ROI, with the majority clustering in the 10–20% range, because only 25–50% of productivity gains are effectively converted into business value.

The regional opportunity: build self-service governance frameworks so operations teams can own their own tooling, reducing the dependency on data science queues. That shifts the knowledge base: operations professionals become operators of AI, not users of it. It’s a talent calibration and a structural one.

What change do you most want to see in post-trade over the next five years, and what comes next for you?

The change I most want to see is the end of reconciliation as an industry.

Think about what post-trade actually is today: thousands of people across the street checking whether two records of the same event agree. That entire discipline exists because our architectures were built around handoffs (front office to middle to back, bank to custodian to CSD), each with its own copy of the truth. Compressed settlement cycles, T+1 and whatever follows it, are exposing how fragile that is. You can’t batch-process your way to instant settlement.

So the five-year shift I want is from detective to preventive by design: trades born correct, enriched once, with controls embedded in the flow rather than inspecting it afterwards. AI accelerates this, but the honest version is that most of the gain comes from unglamorous work: data standards, golden sources, killing handoffs rather than automating them. The firms that get there won’t just be cheaper; they’ll launch products faster and consume less risk capital. That’s when operations stops being a cost conversation and becomes a growth conversation.

As for what’s next for me, that’s exactly why a UAE challenger bank is where I want to build. The global institutions I’ve run operations for carry decades of architectural sediment; you spend half your energy retrofitting. A challenger here has no legacy to defend. You can design the operating model I’ve just described from first principles: front-to-back flows, preventive controls, right-shored from day one, and a talent engine that builds operators for the work that matters rather than the work that used to matter. And the UAE is the right place to do it: regulatory ambition, capital, and a genuine appetite to leapfrog rather than catch up. In five years, I want to have built the post-trade function that incumbents benchmark against, and the team that runs it better than I do.