As the Chief Product Officer at Dooap and a former controller, Anna Tujunen brings a unique “for AP professionals, by AP professionals” perspective to the world of financial technology. With her deep roots in the Microsoft Dynamics 365 ecosystem and her active involvement in the US e-invoicing sub-group, she bridges the gap between rigid ERP structures and the fluid needs of modern accounting teams. In this discussion, we explore the friction points within standard D365 Finance workflows and how specialized automation can transform administrative bottlenecks into streamlined operational advantages.
The conversation covers the limitations of traditional purchase order matching, the hidden costs of manual administrative processing, and the strategic shift toward machine learning-driven self-billing.
D365 Finance relies heavily on Purchase Order matching for all invoices. How does this rigid structure create friction for utility or legal fees, and what specific bottlenecks arise when teams use blanket orders as workarounds?
The fundamental issue is that D365 Finance is built on the unwavering assumption that every invoice must have a corresponding Purchase Order to be valid. When you try to force a variable utility bill or a legal fee into this box, you create a logistical nightmare for the AP team. Often, teams resort to blanket orders, but these are really just bad workarounds that trigger a cycle of manual intervention. For example, a legal firm might send an invoice for an amount that doesn’t perfectly match the remaining balance on a blanket PO, or the line-item coding doesn’t align with the original setup. This forces an AP clerk to manually adjust the PO, seek a new approval for the variance, and re-route the entire document, essentially negating the “automation” the system was supposed to provide.
Administrative expenses often represent 10 to 30 percent of total invoice volume. Why does standard ERP functionality struggle to automate this specific category, and what are the long-term financial costs of neglecting these recurring patterns?
Standard ERPs struggle because they lack a dedicated logic for expenses that don’t fit the classic procurement mold; they view everything as a one-to-one transaction rather than a predictable pattern. When 10 to 30 percent of your invoice volume is treated as an exception, your staff ends up spending nearly a third of their time on low-value data entry and manual coding for things like electricity or rent. The long-term financial cost isn’t just the salary hours wasted; it’s the missed opportunity for early payment discounts and the increased risk of late fees. Neglecting these patterns keeps the department in a reactive state, where “firefighting” administrative tasks prevents the team from performing the high-level financial analysis that truly drives business growth.
While corporate controls are necessary, over-reliance on PO matching can stifle operational agility. In which specific scenarios is a non-PO approach actually considered best practice, and how do you balance oversight with the need for speed?
There are absolutely scenarios where a non-PO approach is the gold standard for efficiency, particularly with fixed-cost recurring services or highly variable but essential utilities. In these cases, forcing a PO process adds layers of bureaucracy without adding any real security. We balance oversight by shifting the focus from “matching” to “automated approval workflows” based on predefined logic. For instance, if a monthly rent invoice falls within a 2% variance of the contract price, it should skip the manual queue entirely. This allows the organization to maintain rigid control over high-value procurement items while remaining agile enough to keep the lights on without a dozen clicks for every utility bill.
Modern tools use machine learning to recognize recurring patterns for rent or service invoices. How does shifting to automated self-billing change the day-to-day role of an AP professional, and what specific milestones should a company hit during this transition?
Shifting to automated self-billing transforms the AP professional from a data entry clerk into a data steward and exception manager. Instead of typing in invoice numbers, they spend their day monitoring system health and optimizing workflows. A key milestone in this transition is the “Recognition Peak,” where the machine learning model successfully identifies 90% of recurring patterns without human prompts. Another milestone is the implementation of self-billing, where the system generates the invoice based on a contract, eliminating the need to even receive a document. This shift drastically boosts employee morale because it removes the repetitive, “soul-crushing” tasks and allows the staff to engage in more meaningful, professional work.
What is your forecast for AP automation?
I believe we are moving toward a future of “Invisible AP,” where the vast majority of financial transactions are processed, coded, and paid without a single human touchpoint. Within the next few years, the reliance on traditional ERP modules for AP will continue to fade in favor of best-of-breed tools that leverage machine learning to handle the 10 to 30 percent of “difficult” invoices that currently cause so much friction. We will see a shift where the AP department becomes a profit center through hyper-efficient cash flow management and advanced vendor negotiation, all powered by the high-quality data that automation provides. The goal isn’t just to work faster, but to build a system so intelligent that it anticipates the needs of the business before the invoice even arrives.