Transforming payment ops: Ctrl AI Dispute to reboot intake

Today's payments landscape is changing rapidly with the dispute and chargeback mix - both volume and complexity - continuing to rise. New digital payment options (e.g., RTP, P2P) drive customer expectations for speed and transparency as regulatory scrutiny and fraud risks mount. Internal efficiency demands simultaneously illuminate prior piecemeal efforts have not kept up - and will continue to be insufficient moving forward.

Senior banking operations executives need more straight-through-processing (STP) to increase speed, accuracy, and compliance while delighting customers. Headlines tout AI (e.g., Machine Learning, Generative AI, or Agentic AI) as a panacea - promising impressive gains by "simply" eliminating manual work. They are keen to truly understand AI and how to effectively shift disputes to lower cost digital channels. Bankers know they cannot just hope for the best in a regulated industry, especially with numerous stories of AI behaving comically, inaccurately, or even illegally. Add to that a recent report by MIT⁽¹⁾ that finds: "Despite the rush to integrate powerful new models, about 5% of AI pilot programs achieve rapid revenue acceleration; the vast majority stall, delivering little to no measurable impact on P&L."

Transformation ultimately requires reimagining every aspect of the dispute lifecycle with the right advanced AI automation for each situation. In Part I, a case for unified processing was made⁽²⁾. For Part II we will "reboot" and go back to the beginning - providing actionable strategies and insights that use the latest technological levers at intake.

The intake challenge:

The old saying is certainly true in disputes: garbage in - garbage out. Limited digital self-service, static forms, or patchwork systems are still all too commonplace. These restrictions typically stem from AHT concerns or development capacity limits that cannot or will not align digital channels to payment network rule changes (e.g., Visa rules & APIs change twice a year in April and October). This approach is unacceptable in 2025. Separate channel intake approaches not only waste development overhead with duplication but often slow resolution times while increasing compliance risks due to unnecessary back-office intervention. It also confuses customers with radically different experiences. Modern dispute management platforms enable unified intake with increased deferment to digital self-service channels for any payment type.

Steps for success:

1. Understand the existing dispute intake processes across all payment types (e.g., credit cards, debit cards, and direct debits like ACH, as well as the rapidly growing RTP/P2P). Determine under which conditions your bank is willing - or not - to offer dispute intake across payments and channels. Identify inconsistencies in both data capture and the customer experience. Prepare for LARGE volume spikes by making disputes "easy" and ensure your E-2-E process can absorb a 20-30% volume increase.
2. Recognize when channel switching or "friction" makes sense to minimize risk for the bank or deliver the best customer experience. Deferring some disputes from digital to employee assisted channels often makes sense due to mobile app screen sizes and the rise in "friendly fraud." By using a combination of machine learning AI that predicts outcomes based on past experiences and business rules that automate actions, banks can effectively manage channel jumping or gaming the system. That lets customers seamlessly move between channels with context, so they do not have to start over while allowing banks to minimize risk exposure by not providing provisional credit too soon.
3. Figure out a "crawl, walk, run" strategy that will work at your bank. An incremental approach allows you to maintain control and monitor roll-out performance so that you remain compliant, avoid overwhelming operations, and keep customers satisfied. Because tech stacks, release schedules, and recency of investments vary widely by bank, determine which combination of channels, payments, and dispute types - in which order - build on each other to provide efficiency gains and minimize risks. Start small, have an end state vision, and be prepared to learn and adjust your plan over time. As an example, you may start with a single non-fraud debit card transaction in online banking. The next step could be enabling the same functionality in your mobile banking app or expanding within online banking to allow a single fraud transaction claim.
4. Set customer expectations at intake to avoid follow-up calls later. Consumer protection regulations (e.g., Reg E in the US, Section 75 in the UK) and payment networks have specific rules for dispute processing. Customers expect rapid resolution, especially in digital channels, but the final resolution often takes weeks or months. Using those regulations and rules along with internal bank policies you should provide the customer with a clear outline of immediate actions (e.g., provisional credit decision) as well as the expected final resolution date. Reaffirm at intake that the bank is advocating for them and network rules drive the process and timing. Finally, point them to digital self-service options to check their claim status at any time.
5. Target simple or "dumb" intake workflows for each dispute type appropriate for each channel. These "static form" approaches with fixed or limited questions without appropriate validations often fail to capture information required by payment network rules that drive post contact outreaches - and increase error rates, drops in service levels. Intake information must be adaptable and embedded directly into the process to gather all the relevant information at the first point of contact. There are common dispute questions, but there are also major differences between payments. At a minimum, your dispute management solution should be driving dynamic intake questions that align to the payment type. However, the solution moving forward for intake is…
6. Deploy agentic AI to automate intake in a natural, conversational way⁽³⁾. Traditional chatbots, while improved by recent AI advances, are still often rigid and remain siloed within a single channel (see your work on #1 above). Conversational AI agents on the other hand can capture disputes the way customer talk - not how network regulations flow. A customer can describe the issue naturally: "I never received the T-shirts I ordered from XYZ Co a month ago for $50." The agent will retrieve/confirm the transaction details, match the information provided to the network rules, and only ask the customer for the required remaining network questions - if any. The agent can then complete the chargeback and execute bank policies around provisional credits and customer notification with no human interaction.
7. Choose best-in-class predictable AI technology to avoid inconsistent outcomes and a lack of transparency⁽⁴⁾. Rather than deploying prompt-based reasoning at runtime, predictable AI effectively combines AI's power with structured workflows and processes. Predictable AI agents operate with built-in governance controls that provide complete visibility into decision-making processes and ensure auditability for compliance - say goodbye to "hallucinations"! Finally, for work as complex as disputes, make sure the technology is "up to the task." Pegasystems was just named a Leader in the 2025 Forrester Wave™ for Digital Process Automation⁽⁵⁾ which drives our market leading Smart Dispute Agentic Automation application⁽⁶⁾.

The future of dispute and chargeback management is intelligent, automated, and adaptive. Senior bank operations leaders can transform for the future; it starts with "clean and complete" AI assisted intake for self-serve while creating the robust foundation required for automation. The benefits are compelling: increased STP, faster resolution times, decreased training, lower operational costs, reduced compliance risk, and a dramatically improved customer experience. As the payments and technology continue to evolve, those who embrace these advanced capabilities will lead the market in efficiency, agility, and trust. It's time to reboot!

Reference links:

1. https://fortune.com/2025/08/18/mit-report-95-percent-generative-ai-pilots-at-companies-failing-cfo/
2. https://www.pega.com/insights/articles/transforming-payment-operations-part-i-unified-dispute-management
3. https://www.pega.com/insights/articles/redefining-self-service-enterprise-workflows-meet-predictable-ai
4. https://www.pega.com/predictable-ai
5. https://www.pega.com/forrester-dpa-2025
6. https://www.pega.com/industries/financial-services/smart-dispute