Kush Singh

Boundary Shift: Operations vs Quality in Payment Screening - In payment screening today, Operations teams process alerts and transactions, while Quality or Risk teams review accuracy through sampling, audits, and post-processing checks. This structure made sense when detection relied heavily on manual review and retrospective validation. With AI integrated into screening — anomaly detection, pattern recognition, and real-time risk scoring — quality signals can now surface during the transaction review itself rather than after the case is closed which makes the process more efficient and agile. The traditional QA role shifts from manual case checking to overseeing model accuracy, bias monitoring, alert tuning, and ensuring regulatory alignment. As we evolve with AI-Driven Screening (at a Level1 phase only) investigators and operations analysts become AI-augmented decision-makers who act on real-time intelligence such as risk indicators, historical behavior patterns, and contextual alerts. Instead of waiting for QA validation cycles, decisions are supported instantly by AI insights. Today In this evolving space the teams like Business Intelligence or Process Excellence function becomes very critical as they focus on interpreting AI outputs, identifying systemic alert patterns, improving screening logic, and ensuring the AI models continue to align with regulatory expectations and evolving financial crime risks.

The Evolution of the Continuous Improvement Specialist in an AI-Enabled World I want to speak specifically about the Continuous Improvement (CI) Specialist — a role I know well! Today, a CI Specialist spends a disproportionate amount of their time on what I call "the framework of problem-solving" · Gathering data · Mapping processes · Building dashboards · Formatting A3s · Running repetitive root cause analyses The are valuable, but not irreplaceable. AI is already beginning to absorb this framework. The data collection, basic process mapping, standard reporting, template-driven DMAIC documentation, and surface-level pattern recognition. An AI tool trained on historical process data can do this in minutes. What a CI Specialist of future will be defined is by their ability to ask the question the data never thought to ask. This is a irreplaceable skill. · The ability to walk a Gemba (operations floor) · Feel the cultural tension in a room · Notice the workaround that no sensor captures. Not that AI cannot stand on a shop floor in Bengaluru and understand why the night shift behaves differently from the day shift. A seasoned CI professional can. The career progression will look like this: In the early years, the expectation will shift. Entry-level CI roles will require AI fluency as a baseline — using tools to accelerate analysis, simulate process changes, and generate improvement hypotheses. Those who cannot will not progress. But those who only rely on AI will move fast, because they will produce competent but culturally thin, contextually shallow work — the "average" that the shrinking mean produces. At the senior level, the most valuable CI leaders will be those who can design the questions AI is trained to answer — shaping how AI tools are deployed inside the organization, what data gets captured, what problems get prioritized, and crucially, what human judgment must never be delegated to an algorithm. The capability that defines every stage of progression: The ability to grow your own knowledge — not just retrieve it. AI can give you an answer. Only a thinking practitioner can know whether it is the right answer for this problem, this culture, this moment. The CI Specialist who thrives in an AI-enabled world is not the one who uses AI the most. It is the one who uses AI to go deeper.

Traditionally, in payment processing or alert adjudication, the role is built around high-volume alert handling, recall of regulatory knowledge, and manual investigation across multiple systems. This is typically a layered decision-making process, meaning the adjudication decision goes through a four- to six-eye check before the payment can be released. Today, AI penetration in this space is still primarily at Layer 1, where the complexity is relatively low, while Layer 2 and Layer 3 exist to validate the decision. Since AI is now being embedded in screening utilities to provide alert adjudication and risk probability scoring, the value proposition of the Layer 2 or Senior Investigator (L3) becomes even more significant. In fact, their role becomes more critical in interpreting AI outputs, challenging model assumptions, and making defensible risk decisions. With this new operating model, hiring criteria should evolve to prioritize analytical judgment, risk contextualization, and the ability to collaborate effectively with AI systems, rather than focusing only on operational throughput or years of manual investigation experience. New hires will increasingly function as risk analysts and AI validators, rather than just transaction processors. Therefore, recruitment should assess how candidates think through ambiguous alerts, identify potential blind spots in AI models, and explain decisions that may override AI recommendations. In summary, the hiring model should shift from asking, “Can this person process alerts at different levels?” to “Can this person detect when the AI is wrong and make a defensible decision?” This shift ensures that organizations gain the full benefits of AI augmentation while maintaining strong compliance, integrity, and operational resilience.

In the banking world (payment processing), AI is increasingly used to recommend repair actions, prioritize investigations, and suggest risk outcomes. While the AI provides decision support, the final action typically remains with the operations analyst. This shift fundamentally changes how performance should be measured. If individuals continue to be evaluated only on traditional productivity metrics such as turnaround time or case volumes, it can unintentionally drive two harmful behaviors: · Blind acceptance of AI recommendations to maintain speed · Resistance to AI usage due to fear of accountability. Therefore, performance expectations must evolve to measure decision quality, risk awareness, and responsible AI usage, rather than only output volume. By embedding responsible AI behaviors into performance expectations, organizations can sustain both human expertise and technological advancement while maintaining regulatory and business integrity. Incorporate Learning and Adaptability Indicators Since AI systems evolve continuously, performance measurement should reward employees who adapt effectively. Examples include: 1. Participation in AI feedback and calibration exercises 2. Contribution to identifying AI errors or drift 3. Ability to handle complex exceptions beyond AI capability Practical Performance Framework for AI-Supported Processes A balanced performance model should combine 3 evaluation pillars: Operational Efficiency - Measures turnaround time and throughput while maintaining defined quality. Responsible AI Interaction Measures appropriate usage, overrides governance, and provides feedback contributions. Learning and Improvement Contribution Recognizes employees who help refine AI capabilities and identify improvement opportunities.

In a banking world, particularly payment operations (SWIFT MT / ISO 20022 flows) AI is increasingly used to flag payments, recommend repair, and adjudicate risk at the first level for sanctions false positives. Now since this is a high-risk and a high-volume process where a minor drift directly impacts business outcomes such as SLA breaches, regulatory exposure, customer dissatisfaction, and operational cost it is of utmost importance to monitor the outcome of AI ensuring that that the output is in line with the rules embedded. AI cannot be treated as a “set-and-forget” capability. Instead, we would need a closed-loop operational governance model with clear ownership and actions keeping the objectives simple but non-negotiable. Some of the key indicators that we I would keep a tack of to ensure if they slip and AI is not doing its job regardless of accuracy scores. Some of the key parameters that I would look at are: · Payment repair rate · Hit rate · Customer query rate · Customer satisfaction score If these metrics degrade, AI drift is already hurting the business even if model precision appears stable.