Accounts payable is entering a new era. After years of focusing on digitization and process automation, organizations are now applying artificial intelligence to improve decision-making, accelerate workflows, and transform AP into a more intelligent, data-driven business function.
This article series is based on the findings from State of AP 2026: AI Rising, a new research report sponsored by Medius and xSuite. Drawing on the perspectives of 194 accounts payable, P2P, and finance leaders, the research explores how organizations are adopting AI, where they are realizing the greatest value, the operational challenges they continue to face, and the capabilities that distinguish top-performing AP organizations.
Throughout this series, we’ll examine the report’s most important findings, including the current state of AP, emerging AI use cases, automation trends, AI readiness and maturity, and the strategies leading organizations are using to prepare for more autonomous finance operations. Whether your organization is beginning its AI journey or expanding existing initiatives, these articles will provide research-backed insights to help guide the next phase of AP transformation.
In Part Ten of our series, we examine how AP has made significant progress automating predictable, rules-based work, while more complex processes such as exception management, cash management, and supplier onboarding remain heavily manual. This creates an opportunity for AI to extend automation into more judgment-intensive activities and pave the way for increasingly autonomous, agentic workflows. However, data quality, system integration, and workforce AI skills remain the primary barriers to scaling AI effectively across AP.
Automation Levels Across the AP Lifecycle
AI and automation are distinct technologies, but they are closely connected. Organizations with more highly automated AP processes are generally better positioned to extend AI across the function. The automation data shows a profession that has made substantial progress digitizing routine, rules-based work while continuing to rely on manual effort for more complex activities.
Invoice approvals, payment scheduling, and travel and expense management represent the most mature areas of automation, reflecting years of investment in well-defined, repeatable workflows. At the other end of the spectrum, exception management, early payment discount capture, cash management, and supplier onboarding remain significantly more manual. These activities require greater coordination, judgment, and cross-functional decision-making, making them more difficult to automate through traditional technologies alone.
The contrast is telling. AP has largely automated the predictable work. The next phase of improvement lies in reducing the manual effort associated with the unpredictable work. Rather than replacing traditional automation, AI has the potential to extend it into processes that have historically depended on human intervention. The opportunity is not simply greater efficiency. It is expanding the range of work that AP can perform intelligently and at scale.
The automation data points toward something larger. As organizations continue to build out their foundational capabilities, improve data quality, and gain confidence with AI in structured workflows, they are quietly creating the conditions that more autonomous AI systems will require. Agentic AI, systems designed to execute connected workflows and complete defined processes with progressively less human intervention, is not a distant concept. For the organizations already deploying AI across multiple functions, the earliest agentic applications are already within reach. The investments being made today are not just improving current performance. They are building the infrastructure for what comes next.
What Is Holding AI Back? The Honest Assessment
If AI adoption is accelerating across the AP market, the barriers slowing it down are worth understanding clearly. They are not technological. The tools are mature, accessible, and improving rapidly. The obstacles are organizational, and for most AP teams they are the same ones that have complicated every major technology transition before this one.
The data is unambiguous about where the friction lives.
Data quality leads by a significant margin, and that result deserves more than a passing acknowledgment. It is not simply the most common barrier. It is the upstream constraint that makes every other barrier harder to overcome. AI systems learn from data. They match, code, route, and flag based on data. When supplier records are incomplete, invoice coding is inconsistent, and transaction histories are fragmented. AI outputs reflect those problems directly. Organizations that have invested in data governance have not just improved their reporting. They have built the foundation that makes AI reliable enough to trust and scale.
Integration compounds the data problem. AP data rarely lives in one place. It moves across ERP platforms, invoice automation solutions, payment systems, and procurement applications. AI delivers its greatest value when those systems are connected and information flows cleanly between them. When they are not, AI becomes a point solution operating on a partial picture, which limits both its accuracy and its organizational credibility.
The skills barrier points to something equally important. AP professionals do not need to become data scientists. But they do need to understand what AI is doing within their processes, when its recommendations can be trusted, and where human judgment remains essential. The organizations making the fastest progress with AI have invested in that fluency deliberately, treating it as a change management priority rather than an afterthought to implementation. The technology can be deployed in weeks. The confidence to use it well takes longer, and it cannot be rushed.
Access this exclusive research today by downloading the State of AP 2026: AI Rising report now.
