The shift from AI adoption to systemic reorganisation in finance automation

For years, finance software makers sold chief financial officers a seductive idea: add smarter automation and the back office would begin to run itself. In practice, many companies have discovered that better models do not automatically produce better outcomes. The problem, increasingly, is not whether artificial intelligence can read invoices, spot anomalies or classify transactions, but whether the wider finance operation is organised in a way that lets those tools work across the w...

That shift in thinking is now at the centre of a broader push toward what some vendors and consultants call touchless finance or autonomous finance. Deloitte has described the concept as “Lights Out Finance”, a model in which AI and machine learning handle routine work end-to-end, leaving people to focus on exceptions, strategy and relationships. Oracle has likewise argued that CFOs are moving toward a more automated operating model, while Workday has pointed to rising use of AI for invoice processing, account reconciliation and data entry.

But the reality inside many enterprises remains messier. PYMNTS reported that while invoice capture accuracy and data extraction have improved, many finance teams have simply bolted AI on top of older, fragmented workflows. The result is a process that may begin more efficiently, yet still breaks down when documents move between systems, teams and approval chains that do not share consistent data or rules.

Michael Younkie, vice president of product management at Billtrust, told PYMNTS that finance teams continue to wrestle with “inconsistent and incomplete data structures, bad data, dirty data” as well as legacy ERP systems with limited accounts receivable API capabilities. That matters because even strong AI performance at the front end cannot fix what happens downstream if the underlying architecture is disjointed.

Research cited by PYMNTS Intelligence suggests many executives now understand this distinction. In its Enterprise AI Benchmark Report, 71% of leaders at companies with annual revenue of $1 billion or more said organisational readiness, rather than AI itself, was the main constraint on performance. Only 11% pointed to the technology as the key barrier. PYMNTS Intelligence also found in December that 66% of accounts payable teams had seen manual workloads increase over the previous year, underlining how stubborn the operational burden remains.

The implication is that the next phase of finance automation is less about adding more intelligence at isolated points and more about redesigning the workflow itself. Instead of treating invoice capture, validation, approval and payment as separate events, the emerging model sees them as connected states within a single system. In accounts receivable, the same logic applies to cash application, collections and dispute resolution.

That is why the most ambitious finance leaders are now asking a different question: not whether a tool has AI features, but whether the entire invoice or cash cycle can run autonomously from start to finish. The answer, for many organisations, still depends on standardising data, simplifying hand-offs and rethinking processes that were built for a manual era.

The destination, at least in theory, is touchless finance. In that model, machines handle the routine work while humans step in only where judgement, negotiation or oversight is needed. The promise is not merely faster processing, but a finance function that is less reactive, less fragmented and more strategic. What is changing now is not the aspiration, but the recognition that reaching it will require rewiring the system, not just upgrading the software.

Source: Noah Wire Services