J.P. Morgan slashes keystrokes with payment automation
J.P. Morgan Payments has reduced 13 billion manual keystrokes annually by automating the paperwork involved in business payments, a shift that addresses one of the most persistent inefficiencies in corporate finance. The sheer volume of manual data entry—once a necessary but time-consuming part of payment processing—has long been a bottleneck for businesses, particularly those dealing with high transaction volumes. Each keystroke represented not just a moment of human labor but also a potential point of error, where misread numbers or misplaced decimal points could cascade into reconciliation delays or financial discrepancies. By eliminating this step, the bank has effectively removed a layer of operational risk while freeing employees from repetitive tasks that offered little strategic value.
The bank’s lockbox network handled approximately 480 million checks and payment documents in 2025, a figure that shows the scale of paper-based transactions still flowing through corporate accounts. Each envelope processed by the network contained a mix of invoices, remittance slips, handwritten notes, and other attachments, all of which varied in format, legibility, and complexity. Some documents arrived folded multiple times, while others were stapled or paper-clipped together, creating physical obstacles that once required manual intervention to separate and sort. The diversity of these materials meant that no two envelopes were identical, forcing processors to adapt to each new variation—a challenge that automation has only recently been equipped to handle.
AI interprets documents while robots manage physical tasks
The bank overhauled its lockbox platform in 2020, integrating AI into its workflows in a way that transformed how unstructured data is processed. After scanning, computer vision algorithms analyze the layout of each document, identifying fields such as payment amounts, invoice numbers, and vendor details. Machine learning models then cross-reference this data against predefined business rules, flagging discrepancies like mismatched totals or missing signatures. The introduction of large language models has further expanded the system’s capabilities, allowing it to interpret handwritten notes, decipher ambiguous instructions, and even resolve exceptions that previously required human judgment. For example, if a remittance slip references an invoice number that doesn’t match the attached document, the AI can now suggest potential corrections based on historical patterns or contextual clues.
The platform handles over 4,000 document types with 99.999% accuracy in data extraction and rule validation, a level of precision that manual processing could never consistently achieve. In 2025, J.P. Morgan introduced robotics to the physical workflow, deploying machines that perform tasks once reserved for human hands. These robots open envelopes, extract checks and invoices, unfold documents, organize paperwork, and prepare everything for AI processing. The integration of robotics addresses a critical gap in automation: while AI excels at interpreting data, it cannot physically manipulate paper. By combining the two technologies, the bank has created a seamless pipeline where documents move from mailroom to digital system without human intervention.
“By investing in robotic and AI technology to improve our lockbox operations, we are automating the most labor-intensive tasks of the process, freeing our team to focus on more complex, higher-value decision-making,” said Michelle Conklin, Head of Receivables and Public Sector at J.P. Morgan Payments. The shift has allowed employees to redirect their efforts toward activities that require analytical thinking, such as resolving complex payment disputes, optimizing cash flow strategies, or identifying trends in client payment behavior. The bank has also reported a reduction in processing errors, which translates to fewer reconciliation delays and lower operational costs for clients.
Robots were initially tested at one location, where they processed a subset of the bank’s lockbox volume to assess performance and identify potential issues. Following this successful trial, the technology is being refined and will be reintroduced later this summer as part of a gradual expansion. The phased deployment approach allows the bank to monitor performance in real time, ensuring that any adjustments are made before scaling the solution to additional facilities.
Related: How ATM Security Enhances Customer Trust and Confidence
Checks remain common, but the manual effort around them is declining
Checks still account for 25% to 26% of U.S. business-to-business payments, a share that has remained stubbornly consistent despite the rise of digital alternatives. The persistence of checks is not due to a lack of innovation in payment technology but rather the inertia of established business practices. Many companies, particularly in industries like manufacturing, healthcare, and construction, have built their accounts payable and receivable systems around check-based workflows. These systems often include approval chains, reconciliation processes, and reporting tools that are deeply integrated with paper-based transactions. For these businesses, switching to digital payments would require not just a change in payment method but a complete overhaul of internal processes—a transition that can be costly and disruptive.
The real challenge isn’t the check itself but the manual effort required to convert paper into usable financial data. When a check arrives, it must be matched to the correct invoice, verified against purchase orders, and recorded in the company’s accounting system. This process, known as reconciliation, is critical for maintaining accurate financial records and ensuring that payments are applied to the right accounts. Historically, reconciliation has been a labor-intensive task, requiring employees to manually cross-reference documents, correct discrepancies, and update systems. Errors in this process can lead to delayed payments, strained vendor relationships, or even compliance issues if financial records are not properly maintained. “Moving dollars is only half the story,” according to Conklin. The other half is ensuring payment data is accurate and actionable the moment funds arrive, helping businesses reduce days sales outstanding (DSO), improve working capital, and accelerate reconciliation.
Automation helps companies shorten days sales outstanding (DSO), a key metric that measures the average number of days it takes to collect payment after a sale. A shorter DSO means faster access to cash, which can be reinvested into the business or used to pay down debt. By accelerating the reconciliation process, AI-driven systems reduce the time between payment receipt and data availability, allowing businesses to act on financial information more quickly. This improvement is particularly valuable for small and mid-sized companies, which often operate with tighter cash flow margins and cannot afford delays in payment processing. Additionally, automation reduces the risk of human error, which can lead to misapplied payments or unrecorded transactions—issues that can take weeks to resolve and may require costly audits to correct.
The transition reflects broader trends in payment processing, where AI is making traditional workflows more efficient without requiring businesses to abandon familiar tools. Rather than forcing companies to adopt new payment methods, banks like J.P. Morgan are focusing on making existing processes faster, more reliable, and less labor-intensive. This approach acknowledges that change in corporate finance is often incremental, with businesses preferring to optimize current systems before investing in entirely new ones. AI’s ability to interpret unstructured data—such as handwritten notes or non-standard invoice formats—means that even the most outdated payment methods can be integrated into modern workflows. The technology does not require documents to conform to a specific template, making it adaptable to the wide variety of formats used by different industries and vendors.
Checks have survived for so long because businesses built decades of workflows around them. What’s changing now are the economics of processing. AI has reached a point where it can interpret thousands of document variations, extract meaning from unstructured data, and automate work that previously required human intervention. Going forward, some of the biggest productivity gains across financial services may come from making legacy payment workflows machine-readable.
