Across finance, AI use is clearly shifting from pilot programs to becoming integral in day-to-day operations. What was considered experimental just a year or two ago is now foundational to daily workflows.
According to the AI Momentum Report, 72% of the nearly 800 finance leaders surveyed are using some form of AI in finance or accounts payable (AP), and 82% plan to increase their AI investments over the next 12 months.
Gartner reports that a majority of finance teams are now using AI in some form, while Deloitte’s CFO Signals survey found that 87% of CFOs expect AI to play an important role in finance operations in the near term.
But interest and adoption alone does not equate to meaningful impact. Across finance organizations, there is a growing gap between how AI is being deployed and if it is delivering consistent, measurable results.
Adoption may be widespread, but maturity is uneven
Despite high levels of AI use and adoption, many core operational challenges remain. Manual data entry, high processing costs, and slow approval cycles continue to be cited as persistent issues, even in environments where advanced automation and AI may already be in use.
This pattern is not unique to AP. Across finance more broadly, research from firms like Deloitte and KPMG has found that while AI initiatives are increasing, many fail to scale or deliver meaningful returns within a reasonable time. Accenture, for example, reports that while 86% of executives plan to increase their AI investments, only 32% say they are achieving sustained, enterprise-wide impact.
Our exclusive AI Momentum Report also found that nearly half of organizations are still in pilot or early scaling phases of AI adoption. So in practical terms, AI is present, but not yet fully integrated or optimized for how work gets done.
Execution, not access, is the limiting factor
Finance does not have an access problem when it comes to AI. Rather, it has an execution problem.
Many organizations are introducing AI into environments that are only partially automated. ERP systems are widely used, but integrations with automation platforms are often incomplete, leaving critical manual process steps in place.
At the same time, AI is frequently layered onto fragmented or broken workflows rather than being used to reimagine the way work should get done. This band-aid approach results in some improvement at specific points in the process, but limits reaching true end-to-end autonomy across the invoice-to-pay process.
This helps explain why gains are often incremental rather than transformative.
Why the type of AI matters in accounts payable
Not all AI is created equally. Increasingly, the type of model underpinning a solution is proving just as important as the decision to adopt AI itself. Domain-specific AI trained on highly relevant industry data can offer a more practical path to value.
Gartner notes that these models “facilitate faster time to value for AI projects” and recommends evaluating off-the-shelf, domain-specific options before building from scratch. Vic.ai, for example, has trained its proprietary model on over one billion invoices, bringing hyper-focused AI to the table for AP teams, where context and precision are critical.
What early success can look like, in practice
Some organizations are already closing the adoption-to-value gap. In a recent webinar, finance leaders from Ancestry described how their approach to AI in AP focused on targeted implementation rather than broad transformation. By reimagining specific workflows with AI — in their case, invoice processing — they were able to reduce manual work quickly and see measurable improvements without overhauling their entire system infrastructure at once.
One key takeaway for Ancestry: the value of AI quickly became visible once it was embedded directly into existing processes. Rather than adding complexity, they reduced operational burden and allowed the AI system to learn over time. This created early wins, which in turn helped build internal confidence and support further adoption.
This pattern (focused use case, fast feedback, incremental scale) is consistent with what the broader industry findings and best practices suggest.
What more effective approaches have in common
Organizations seeing stronger results from AI in AP tend to share a few characteristics:
- They start with clearly defined use cases, focusing on high-volume, error-prone workflows such as data extraction and invoice approvals.
- They prioritize integration, ensuring AI works within existing systems rather than alongside them.
- They measure performance beyond surface-level efficiency, evaluating whether manual work is actually being reduced.
- And importantly, they treat AI as an operational capability, not a one-time deployment. Implementation is iterative, requiring ongoing refinement and alignment with business outcomes.
These differences are less about technology and more about adoption, execution, and change management.
From adoption to impact
While concerns around AI accuracy, security, and oversight remain, confidence tends to build when systems consistently perform within real workflows and align with existing controls. At the same time, finance teams are beginning to redefine what success looks like. The focus is shifting beyond efficiency toward creating capacity for more strategic work — analysis, oversight, and decision-making — while still maintaining appropriate levels of human involvement.
AI is now firmly established in AP and finance, and adoption is no longer the question. The challenge ahead is ensuring that it delivers consistent, reliable, and scalable results. Organizations that are succeeding are not simply adopting AI; they are implementing it effectively, aligning systems, processes, and teams around clear expectations and outcomes.
In finance, the real advantage is no longer access to AI, but the ability and knowledge to make it work.
About Vic.ai
Vic.ai is an AI-native platform purpose-built for the real-world complexity of enterprise accounts payable. We partner with finance teams to operate AP with greater accuracy, control, and confidence — without adding operational burden or headcount. Unlike rules-based or template-driven systems, Vic.ai’s autonomous technology learns continuously, delivering consistent, explainable results across multi-entity environments and high invoice volumes. With Vic.ai, finance organizations gain more capacity, clearer visibility, and the ability to scale as transaction volumes grow. Today, Vic.ai customers have processed over one billion invoices and achieved nearly $200 million in cost savings through autonomous AP. Vic.ai is jointly headquartered in Miami, Florida, and Oslo, Norway. Learn more at www.vic.ai.
