Our recent On-Demand Webinar: How to Utilize AI in AP for Fast, Guaranteed ROI, hosted in partnership with the Executive Leaders Network, featured special guest Meng Lee, Senior Analyst at Forrester, and Leigh Celones, Director at Basware.
CFOs are ready to invest in AI - but don’t know where to start
Basware’s recent survey of over 400 global senior finance professionals, including CFOs, reveals a strong appetite for AI, but also a lack of clarity around how to apply it most effectively.
- 75% of finance leaders plan to increase investment in AI, but don’t know where to begin.
- Most expect to see a return within 12 to 18 months—and risk losing project funding if they don’t.
- Their main goals are to reduce costs (76%) and increase process efficiency (64%).
“We're past the hype cycle,” said Leigh Celones. “CFOs aren’t asking whether to use AI—they’re asking how to use it in ways that are truly additive to the business.”
AI is delivering strong returns for accounts payable
Survey data shows that accounts payable is one of the top two areas where CFOs expect the highest return from AI, alongside FP&A. Why AP?
- It’s rich with manual, repetitive tasks.
- 85% of CFOs identified AP as a high-value area for AI application.
- AP involves high volumes of invoice documents and multiple touchpoints by team members, making it a natural candidate for automation.
- AP processes are clearly defined and data-rich, meaning AI can make fast, measurable improvements.
- Quick wins in improving accuracy, efficiency, and turnaround time give finance teams an early use case to build momentum.
According to Meng Lee, AP is already in the “green quadrant” of Forrester’s AI maturity heatmap, indicating both the technology and adoption readiness are in place.
Basware empowers AP teams with embedded AI at every level
Basware’s InvoiceAI framework and roadmap bring AI into the product where and when it delivers the most value across the whole invoice lifecycle, providing end-to-end enhancement of your AP process. Leigh shared several real-world examples:
- AI agents, including the AP business agent, assist users with practical, user-friendly recommendations enabling them to solve problems or answer questions without having to contact the AP department. Included in Basware Insights Pro, Basware’s AP Data Agent integrates generative AI, allowing teams to ask natural language questions like “What’s our invoice volume by supplier?” and “Where are we missing early payment discounts?” This makes data exploration faster and more accessible to non-technical users.
- AI also helps enrich data by analyzing historical trends across suppliers and spend categories, helping to uncover hidden patterns, optimize spend and support strategic decision making.
“This is human-assisted AI,” Leigh explained. “AI presents options, but the user stays in control.” It’s important for AI to enhance, not replace finance professionals. InvoiceAI is designed with this in mind, supporting human judgement with insights, reducing complexity, and enabling faster, more confident decisions.
The huge potential of agentic AI
Meng and Leigh agreed that 2025 marks a tipping point for agentic AI. These tools not only suggest actions, but execute them on behalf of users. These agentic tools operate within clearly defined guardrails. They’re designed to automate routine steps, without overriding human oversight.
Basware has already launched two agentic tools, including an AP business agent that:
- Surfaces historical data and context.
- Helps users prioritize incoming invoices.
- Reduces time spent triaging low-risk actions.
But, full autonomy is still not the goal. “Finance decisions will always require oversight,” Leigh emphasized. “AI should support, not replace, strategic thinking.”
Key risks to watch out for
Meng identified four key risks finance leaders must manage:
- Data privacy and security. Sensitive financial information must be protected at all times.
- Model bias and inaccuracy. Poor training data can produce flawed or discriminatory outcomes.
- Lack of explainability. “Black box” decisions, where it’s unclear how or why an AI system arrived at a specific outcome, undermine trust and compliance.
- Liability. Organizations remain accountable for incorrect actions taken by AI systems.
Recommended safeguards:
- Apply tolerance levels based on invoice value or transaction type.
- Set confidence thresholds for when AI decisions can proceed autonomously.
- Maintain human-in-the-loop controls for exceptions and high-value decisions.
These guardrails help to keep your AP automation strategy secure, transparent, and auditable.
How to measure ROI
Forrester’s ROI framework for AI initiatives includes three key pillars:
- Quantifiable benefits: cost savings, increased throughput, fewer delays
- Investment costs: technology, implementation, and internal resources
- Risk adjustment: factoring in AI limitations, compliance exposure, and macroeconomic volatility
Leigh added that adoption rate is a critical success factor. “Even with the best tools, if users aren’t adopting the AI, the ROI never materializes.”
Practical advice from the experts
- Start with proven AP use cases: Focus on “sense” (AI reading and interpreting data) and “decide” (AI making recommendations). These are mature, low-risk areas where success is well proven.
- Think program, not project: Ongoing improvement, governance, and iteration are essential to long-term AI value.
- Build clear governance: Use thresholds and tolerances to balance automation with control. Demand transparency from vendors.
- Align to strategic goals: Whether your priority is cost savings or process transparency, AI should support, not distract from, your business objectives.
Watch the webinar on-demand
Watch the full On-Demand Webinar: How to Utilize AI in AP for Fast, Guaranteed ROI, hosted in partnership with the Executive Leaders Network, to hear directly from Forrester and Basware on how CFOs are using AI to drive results in AP. You’ll also gain access to Forrester’s AI Adoption Heatmap, Basware’s ROI tools, and much more.