“RIDP Provides A Much-Needed Platform For Controlling Fraud Risk”
As India’s digital payments ecosystem continues its explosive growth, it is also facing a parallel surge in financial fraud. In this exclusive interview, Rajeev Ranjan, Editor at Digital Terminal, speaks with Deepak Chand Thakur, Co-founder & CEO of NPST, about tackling modern payment fraud. Deepak explains how NPST’s AI-driven Risk Intelligence and Decisioning Platform (RIDP) is redefining how payment fraud is detected and prevented
Rajeev: With digital financial fraud surging by 67% in FY25, what do you see as the most pressing risks for acquiring banks and payment aggregators in today’s UPI-dominated ecosystem?
Deepak: The 67% surge in digital financial fraud in FY25 highlights a stark reality. Acquiring banks and payment aggregators are operating in an increasingly complex and high-stakes risk environment, especially within the rapidly expanding UPI ecosystem.
One of the most pressing challenges is the democratization of the acquiring space. What was once the domain of a few large banks is now opening to non-bank entities such as payment aggregators, co-operative banks, and smaller financial institutions. While this shift certainly enhances financial inclusion and innovation, it also introduces systemic vulnerabilities, especially when these new entrants lack the infrastructure or maturity to support real-time, scalable risk management.
Another structural concern is the economic asymmetry of the acquiring business. Acquirers earn only a small fraction of each transaction’s value, yet bear the full liability for fraud, chargebacks, or regulatory lapses. This imbalance means even a modest fraud event can cause outsized financial and reputational damage. When you’re dealing with millions of merchants and sub-merchants, the old playbook i.e., relying on static labels like low, medium, or high risk, is simply not enough. Risk is no longer static; it’s dynamic, fluid, and constantly evolving.
A further layer of complexity comes from the limitations of legacy transaction monitoring systems. While they may flag behavioural anomalies, they often fail to provide contextual intelligence. They might miss crucial early indicators such as:
Sudden spikes in transaction volumes,
Uncommon number of transactions
An increase in refunds or chargebacks,
Quiet changes in merchant metadata or KYC information,
Or reputational signals hiding in digital footprints and media sources.
Without a holistic and continuous risk view, acquirers are left exposed, not just to fraud, but to transaction laundering, regulatory scrutiny, and mounting compliance costs.
At NPST, we believe the answer lies in taking a fundamentally different approach to risk. Our Risk Intelligence and Decisioning Platform (RIDP) is purpose-built to tackle these modern realities. By leveraging AI and big data, RIDP gives acquirers a 360-degree view of merchant behaviour—not just at onboarding, but throughout the entire lifecycle. It ingests and analyses signals from multiple sources—transactional patterns, business metadata, customer sentiment, adverse media, and compliance indicators—to dynamically assess and recalibrate merchant risk.
Rajeev: How does NPST’s Risk Intelligence Decisioning Platform (RIDP) leverage AI/ML to address evolving merchant-level fraud challenges that traditional monitoring systems struggle to manage?
Deepak: The challenge with traditional fraud detection systems is that they are designed around static rules. They’re good at spotting what’s already known, but far less capable when threat vectors are agile and more nuanced. And static systems, no matter how well-calibrated, are struggling to keep up.
A core issue is something industry insiders refer to as model drift: the gradual loss of accuracy in fraud detection tools as fraud tactics evolve faster than the systems meant to stop them. This drift doesn’t just weaken protection, it floods fraud teams with false positives and allows actual threats to slip through the cracks.
NPST’s Risk Intelligence Decisioning Platform (RIDP) addresses these challenges by fundamentally rethinking how merchant-level fraud is monitored and managed. At its core is a high-performance AI/ML engine that helps acquiring institutions move beyond surface-level monitoring to a much deeper, more adaptive understanding of merchant risk. From onboarding through to live transaction analysis, RIDP continuously interprets behavioural signals that would escape conventional models.
Crucially, the platform operates on what NPST calls a “Daily Adaptive” framework. Its supervised learning models retrain every day using fresh data streams ensuring that the platform’s risk intelligence evolves in lockstep with tactics deployed by fraudsters. Currently the platform supports use cases such as detecting anomalies in merchant transacting patterns, identifying compliance weaknesses via dynamic scoring, and flagging fraudulent chargebacks.
The outcome is a smarter, more responsive system — one that allows acquiring banks and payment providers to manage risk at the merchant level more effectively and meet growing regulatory expectations around explainability and financial integrity. At a time when digital payments are accelerating and scrutiny on financial institutions is intensifying, RIDP provides a much-needed platform for controlling fraud risk.
Rajeev: Can you explain how RIDP’s real-time anomaly detection and behaviour-based merchant risk scoring work in practice to proactively prevent fraud?
Deepak: Every merchant and merchant category has a distinct transactional DNA. This includes patterns in transaction volume, frequency, geographic footprint, customer demographics, and even time-of-day preferences. NPST’s Risk Intelligence Decisioning Platform (RIDP) uses this data to establish a dynamic baseline for each merchant. It’s not just looking at whether a transaction exceeds a threshold; it’s asking whether the behaviour aligns with the pattern for that particular merchant and with the larger category DNA
For example, if a small business that typically handles modest, domestic food delivery transactions suddenly starts processing a flurry of high-value sales, RIDP recognises the behaviour as anomalous—because it's out of character - and flags the same for identification. This context-aware detection is where conventional rules-based systems often fall short.
The platform’s intelligence runs deeper than just spotting outliers. It can detect anomalies that conventional systems miss—like a spike in chargebacks, or a rise in consumer complaints, an uptick in refund cycles, or changes in the timing of transactions. That anomaly is flagged for immediate review or automated action before it escalates into chargebacks, consumer disputes, or fraud loss.
For example, RIDP has uncovered instances where malicious actors conducted numerous ₹10 transactions and initiated disputes. On the surface, the losses were negligible. However, non-compliance with regulatory response timelines led to cumulative penalties far exceeding the value of the transactions. RIDP flagged this emerging pattern early, enabling acquirers to intervene and mitigate the risk. Even as fraudsters adapted, changing transaction amounts or timings to evade detection, RIDP’s adaptive learning continued to detect behavioural anomalies and adapt counter fraud techniques.
In essence, RIDP provides a predictive counter-fraud intelligence layer that adapts in real-time, detects outliers, and empowers acquirers to take pre-emptive action. The result: acquirers benefit from lower fraud losses, better regulatory compliance, and a healthier, more trustworthy acquiring ecosystem.
Rajeev: What role does automated onboarding due diligence play in minimizing merchant fraud risks?
Deepak: Automated onboarding due diligence plays a pivotal role in reducing merchant fraud risks by significantly improving the accuracy, speed, and depth of merchant verification right from the beginning of the relationship. Know Your Business (KYB) and due diligence checks are essential to underwriting. They enable acquirers to properly assess merchant legitimacy and screen out potentially fraudulent or high-risk applicants early in the process.
While there’s no one-size-fits-all standard, effective due diligence today often includes validating business documents, performing geolocation and digital footprint analysis, conducting live video verification, and leveraging third-party data sources to examine incorporation details, leadership structures, operational history, and prior transaction behaviours.
However, many payment service providers traditionally treat onboarding as an annual exercise, performing manual compliance checks upfront and then failing to monitor merchants continuously. Once a merchant is approved, there’s often limited ongoing oversight, creating a blind spot that fraudsters can exploit by altering their business practices post-onboarding.
This approach exposes them to risks because merchants can change their business models or sales tactics post-onboarding, sometimes engaging in suspicious or fraudulent activities.
The challenge is that manual business and compliance checks are costly and time-consuming, especially in a low-margin acquiring industry. KYB costs for acquirers include investments in technology, third-party data sources, skilled compliance personnel, and ongoing audits, all of which add significant operational expenses.
NPST’s Risk Intelligence and Decisioning Platform (RIDP) transforms onboarding by embedding automated, intelligent checks right from the outset and maintaining persistent risk monitoring throughout the merchant lifecycle.
At onboarding, RIDP’s Risk Check API analyses a wide range of data points, including verified business information, global watchlists, adverse media, and digital footprint signals. Each merchant is assigned a dynamic risk rating that continuously adjusts based on ongoing transactional behaviour, compliance status, and customer sentiment.
Rather than a static gatekeeping step, RIDP’s continuous monitoring flags anomalies like sudden transaction spikes, unusual geographic activity, or mismatches between declared and observed business categories. This real-time vigilance enables acquirers to act proactively, flagging accounts for review, or even blocking high-risk activity, pausing settlements, or adjusting transaction limits before minor red flags escalate into major fraud incidents.
Continuous vigilance helps acquirers reduce fraud risk, maintain regulatory compliance, and safeguard their ecosystem, while managing the costs and inefficiencies traditionally associated with KYB.
Rajeev: Looking ahead, how do you see technology like RIDP contributing to a safer and more resilient UPI and digital payments infrastructure in India over the next 3–5 years?
Deepak: Safety and trust will always be at the core of digital payments. Over the next 3–5 years, we RIDP will continue to play a role in helping acquirers deploy predictive counter-fraud measures.
As fraud typologies evolve, financial institutions, fintechs, and regulators must work together to proactively address emerging threats. RIDP is designed with that future in mind.
We’re focused on expanding the platform’s capabilities, ingesting more diverse data sets and continuously improving the precision of our AI/ML models to help acquirers and issuers navigate an increasingly complex fraud landscape.
Crucially, we anticipate a shift towards greater collaboration and shared intelligence across the ecosystem. One transformative development could be the creation of a centralized bad-actor repository — a shared case management system accessible to trusted industry stakeholders. This would enable faster identification of repeat offenders and systemic fraud patterns. RIDP has been architected to support such a collaborative model, providing the real-time decisioning infrastructure needed to operationalize shared insights at scale.
In this connected future, RIDP will serve as a vital enabler, helping the industry stay ahead of fraud, and reinforcing long-term trust in the digital payments ecosystem.
𝐒𝐭𝐚𝐲 𝐢𝐧𝐟𝐨𝐫𝐦𝐞𝐝 𝐰𝐢𝐭𝐡 𝐨𝐮𝐫 𝐥𝐚𝐭𝐞𝐬𝐭 𝐮𝐩𝐝𝐚𝐭𝐞𝐬 𝐛𝐲 𝐣𝐨𝐢𝐧𝐢𝐧𝐠 𝐭𝐡𝐞 WhatsApp Channel now! 👈📲
𝑭𝒐𝒍𝒍𝒐𝒘 𝑶𝒖𝒓 𝑺𝒐𝒄𝒊𝒂𝒍 𝑴𝒆𝒅𝒊𝒂 𝑷𝒂𝒈𝒆𝐬 👉 Facebook, LinkedIn, Twitter, Instagram
