In the high-stakes theater of modern corporate finance, a new mandate has emerged: deploy agentic AI or risk falling behind. However, a sweeping new report from Avalara, Agents of Change: How the Race to Deploy AI Agents is Outrunning Financial Governance, reveals that this pursuit of competitive advantage is coming at a significant cost. Finance leaders, under immense pressure to justify their AI investments, are increasingly prioritizing the velocity of implementation over the foundational elements of governance and accountability.
The survey, which polled over 1,500 CFOs and senior finance leaders across the U.S., UK, India, and Australia, paints a sobering picture of an industry in a state of high-speed experimentation. While the potential for transformative ROI is clear, the path to achieving it is currently paved with operational risks and a fragmented approach to oversight.
The Pressure Cooker: Why Finance is Racing
For the modern CFO, the implementation of agentic AI—autonomous systems capable of performing complex financial tasks—has shifted from an "innovation project" to a core strategic imperative. According to the Avalara data, more than 90% of senior finance leaders are facing moderate to significant career pressure to prove that these investments are delivering tangible returns.
The Metrics of Urgency
This pressure is manifesting in a lopsided focus on speed. When asked to characterize their organization’s approach to AI deployment, 29% of respondents admitted their focus was "entirely on speed," while an additional 41% said it was "mostly on speed." Perhaps most concerning is that only 7% of surveyed leaders indicated that governance was being prioritized over speed.
This data suggests that the C-suite is caught in an "AI proof gap," where the fear of being perceived as a laggard outweighs the long-term risks associated with unvetted autonomous workflows. The mandate from the boardroom is clear: show the ROI, and do it now.
Chronology: The Evolution of the AI Finance Landscape
To understand why governance is being sidelined, one must look at the rapid evolution of financial technology over the last 24 months.
- Phase 1: The Hype Cycle (Early 2025): Organizations began experimenting with Large Language Models (LLMs) to automate basic documentation and report generation. During this period, the focus was on identifying low-hanging fruit.
- Phase 2: The Agentic Shift (Late 2025): The industry moved from simple chatbots to "agentic" systems—AI that could not only write text but initiate actions, such as tax filings, invoice processing, and cash flow forecasting.
- Phase 3: The Governance Crisis (Mid-2026): As agentic AI reached a critical mass of deployment, the limitations of early, rushed implementations became apparent. Organizations began experiencing "AI drift," where autonomous agents made suboptimal decisions based on poor data inputs, leading to the current discourse on the necessity of robust governance frameworks.
Supporting Data: The Reality of ROI
Despite the aggressive push, the actual financial returns have been mixed. The Avalara survey found that while 90% of leaders reported some form of ROI, the quality and scalability of those returns are questionable.
The Scalability Bottleneck
- ROI at Scale: Only 38% of respondents described their ROI as being achieved "at scale."
- Limited Impact: A full 50% of respondents described their current AI ROI as "limited."
These numbers suggest that while AI can easily solve individual, low-complexity tasks, scaling those successes across a complex, global finance function is proving to be a much more difficult hurdle. The "limited" nature of these returns is often a direct result of implementing AI on top of existing, inefficient workflows rather than reimagining those workflows to accommodate autonomous agents.
Official Perspectives: The Case for Caution
Hugo Sarrazin, CEO of Avalara, has been a vocal proponent of slowing down to speed up. In a recent statement accompanying the report, he cautioned against the dangers of "velocity without validation."
"Finance leaders are right to move quickly to capitalize on agentic AI opportunities," Sarrazin noted. "But speed without accountability creates new forms of risk, and speed without rethinking workflows limits ROI. The organizations that realize the greatest value from AI won’t simply deploy more agents. They’ll leverage agents with trusted data, governed workflows, and clear controls that enable automation with confidence."
Industry experts and analysts echo these sentiments. A previous study by Grant Thornton identified that the primary barrier to AI underperformance is not the technology itself, but the "AI proof gap"—a disconnect between what the AI can do and what the organizational governance structure allows it to do safely.
The Human-in-the-Loop Safeguard
Despite the fixation on speed, the survey revealed a surprising, albeit cautious, reality: human oversight remains the industry standard. Even in companies moving at breakneck speeds, few are willing to hand over the keys to the kingdom.
The Autonomy Gap
The survey profiled seven distinct business processes and found that in six of them, fewer than 15% of leaders are allowing AI to act fully autonomously. The outlier is financial planning and analysis (FP&A), where 19% of functions operate without human intervention.
In almost every other instance, the "Human-in-the-Loop" model prevails. AI acts as a recommendation engine, providing data-driven insights that a human finance professional must then verify and approve. This represents a pragmatic approach to risk management, serving as a vital check against the hallucinations and technical errors inherent in current AI models.
Implications: The Future of Financial Governance
The tension between the demand for rapid innovation and the necessity of risk management will define the next decade of finance leadership. As companies move beyond the "experimentation" phase, the implications of the current "speed-first" culture are becoming clear.
1. The Risk of Regulatory Backlash
As AI agents become more deeply embedded in tax, compliance, and financial reporting, regulatory bodies are likely to take notice. If an autonomous agent triggers a massive tax error or a compliance breach, the excuse of "moving too fast" will not suffice for regulators. Organizations that fail to build governance into their AI architecture now will likely face significant penalties in the future.
2. The Shift to "Workflow-First" AI
The future of successful AI deployment lies in re-engineering business processes. Instead of force-fitting AI into current, legacy-heavy workflows, successful firms are beginning to redesign these processes to be "AI-native." This requires a shift in mindset: seeing AI not as a tool that automates existing tasks, but as a catalyst for entirely new, more efficient ways of operating.
3. The Talent Gap
As the demand for AI integration grows, there is a burgeoning need for a new type of finance professional—one who is as comfortable with data science and AI ethics as they are with Generally Accepted Accounting Principles (GAAP). The race for ROI will eventually be won by companies that possess the talent to govern their agents effectively.
4. A Maturation of Expectations
The "limited" ROI reported by half of the survey respondents is a signal that the market is beginning to mature. The initial excitement of AI is being replaced by a more sober, analytical approach to investment. We should expect future reports to show a decline in "speed-at-all-costs" mindsets in favor of a "value-at-scale" strategy.
Conclusion: Balancing Velocity and Integrity
The findings from Avalara serve as a critical wake-up call for the finance industry. While the pressure to innovate is understandable in a competitive global market, the race to deploy AI agents must not come at the expense of the trust and accuracy that form the bedrock of the finance profession.
For CFOs and finance leaders, the path forward is clear: speed is a multiplier, but only when it is directed by robust governance, fueled by trusted data, and guided by human oversight. The organizations that succeed in the coming years will be those that learn to distinguish between reckless haste and calculated momentum.
As the industry navigates this transformation, the focus must shift from how fast we can deploy, to how reliably we can integrate these powerful tools into the complex, interconnected world of global finance. The goal is no longer just to be the first to adopt, but to be the first to build a model of autonomous finance that is as accountable as it is efficient.
