The financial sector is undergoing a profound transformation driven by agentic artificial intelligence, which is proving to be a catalyst for significant returns on investment. While general AI projects yielded an average ROI of 67 percent last year, autonomous agent systems have delivered an impressive 80 percent by autonomously managing complex operational workflows.
This substantial performance gap highlights a crucial need for technology leaders to re-evaluate their automation budget allocations. Agentic AI is moving enterprises from theoretical value to tangible financial gains. Unlike generative tools that primarily summarize information or draft text, these sophisticated agents are engineered to execute detailed processes within predefined parameters and strict approval limits.
Executive pressure is accelerating this strategic shift. A recent study by Basware and FT Longitude revealed that nearly half of Chief Financial Officers face demands from leadership to integrate AI across their operations. However, 61 percent of finance executives admit initial custom-developed AI agent deployments were more experimental than problem-solving, often failing to deliver expected ROI. Traditional AI models frequently generate insights requiring human interpretation, whereas agentic systems embed decision-making directly into workflows, thereby closing the gap between insight and action.
Basware CEO, Jason Kurtz, observed a diminishing tolerance for unfocused experimentation, noting that boards now expect concrete results from AI investments, branding "AI for its own sake" as wasteful.
Accounts Payable: The Proving Ground
Finance departments are increasingly directing these autonomous agents toward high-volume, rule-based environments. Accounts payable (AP) stands out as a primary use case, with 72 percent of finance leaders viewing it as an obvious starting point. The AP process, involving structured data from invoice capture to payment booking, aligns perfectly with agentic deployment models.
Current implementations leverage agents to automate invoice data entry, detect duplicate invoices, identify fraud, and reduce overpayments. These are not theoretical applications but operational tasks where algorithms function with high autonomy when parameters are accurately configured. Success in this domain relies heavily on data quality; systems trained on vast datasets, like Basware's two billion processed invoices, enable context-aware predictions that differentiate legitimate anomalies from errors without human intervention. Kevin Kamau, Basware's Director of Product Management for Data and AI, characterized AP as an ideal proving ground due to its unique blend of scale, control, and accountability.
The Build Versus Buy Dilemma
Technology leaders must strategically decide how to procure these advanced capabilities. The broad term "agent," covering everything from simple scripts to complex autonomous systems, complicates procurement choices. Procurement strategies often split by function. For accounts payable, 32 percent of finance leaders prefer embedded agentic AI in existing software, while 20 percent choose in-house development. Conversely, for financial planning and analysis (FP&A), 35 percent opt for self-built solutions compared to 29 percent for embedded ones.
This divergence suggests a pragmatic rule: if the AI improves a process shared across many organizations, such as AP, integrating a vendor solution is often more efficient. If the AI creates a unique competitive advantage, building in-house is typically the better path. The general principle advises buying to accelerate standard processes and building to differentiate.
Governance as an Enabler
Fear of autonomous errors often hinders adoption, with almost half of finance leaders (46%) unwilling to deploy agents without clear governance. This caution is rational, as autonomous systems demand stringent guardrails, especially in regulated environments. However, the most successful organizations utilize governance not to impede deployment but to scale it. These leaders are significantly more likely to employ agents for complex tasks like compliance checks (50%) compared to their less confident peers (6%). Anssi Ruokonen, Basware's Head of Data and AI, suggested treating AI agents like junior colleagues, advocating for thorough testing and gradual introduction of autonomy while maintaining human oversight for responsibility.
While a third of finance leaders express concern about job displacement, proponents argue that agents shift the nature of work rather than eliminating it. Automating manual tasks frees staff to focus on higher-value activities, moving from task efficiency to operational leverage. This allows finance teams to achieve faster closes and make better liquidity decisions without increasing headcount.
Organizations that extensively use agentic AI consistently report superior returns. Leaders deploying these tools daily for functions like accounts payable achieve better outcomes than those limiting usage to experimentation. Confidence grows through controlled exposure; successful small-scale deployments foster broader operational trust and increased ROI. To replicate the success of early adopters, executives must transition from unstructured experimentation. Data indicates that 71 percent of finance teams with weak returns acted under pressure without clear direction, unlike the 13 percent achieving strong ROI. Success requires embedding AI directly into workflows and governing agents with the same discipline applied to human employees. As Kurtz concluded, agentic AI can deliver transformative results, but only when deployed with purpose and discipline.
This article is a rewritten summary based on publicly available reporting. For the original story, visit the source.
Source: AI News
