Collections Finally Gets Its AI Glow-Up

What is relied upon can often get taken for granted. That’s the position the accounts receivable (AR) function has traditionally found itself in. After all, what’s more important than getting paid?

But despite its bottom-line impact, AR has traditionally been run as a back-office function with all the limited investment and manual accoutrement that implies. That’s all beginning to change. A combination of one part maturing technologies like artificial intelligence (AI), and two parts economic volatility is forcing businesses to do more with less and rethink where money gets left on the table, and for how long.

“The need to chase customers for payment effectively, that’s not new. It’s just more relevant today because of economic pressures,” Dave Ruda, Billtrust’s vice president of software products, told PYMNTS.

Across industries, after all, suppliers are feeling squeezed. Delinquencies are rising not necessarily because customers are unwilling to pay, but because macroeconomic conditions are forcing prioritization.

“We see suppliers of goods and services having to chase delinquency on buyers more frequently,” Ruda said, pointing to “tariffs, economic pressures and interest rates fluctuating up and down.”

The labor-intensive approach many organizations take to AR, frequently operating out of spreadsheets, exporting aging reports from enterprise resource planning (ERP) systems, and manually orchestrating outreach campaigns, is no longer sufficient in today’s world.

From Spreadsheets to Systems of Intelligence

What’s changing is not just the tooling around AR, but the underlying philosophy of collections itself. No longer a reactive function focused on chasing payments, AR is becoming a proactive, data-driven engine that shapes customer interactions and influences cash flow strategy.

“Send an email, make a phone call at the right time to the right person with the right message, and you can get them to do the thing that you want them to do,” Ruda said.

The difference now is that AI can operationalize that logic at scale. Instead of static aging buckets, systems can ingest historical payment patterns, segment customers dynamically and prioritize outreach based on the likelihood of payment. The goal is not just efficiency, but precision.

Collections, after all, are not just transactional; it’s relational.

“Every outreach is a reflection of your brand,” Ruda said, underscoring the need for guardrails as automation expands.

He stressed that next-generation AR solutions, like Billtrust’s own offerings, should be able to answer at least three key questions to optimize collections: “Who should I be talking to? Why am I talking to them? And then what should I say to them?”

This shift is particularly valuable for lean teams. Many companies are growing revenue without proportionally expanding headcount, forcing small collections teams to manage increasingly complex portfolios. AI, in this context, can function as a multiplier.

“We want to give them tools to basically put a data scientist on their shoulder,” Ruda said, describing a system that could help make “every keystroke … 20x more effective.”

How AI Is Forcing a Rethink of Traditional Metrics

As workflows evolve, so must the metrics used to evaluate them. Traditional key performance indicators (KPIs) such as days sales outstanding (DSO) or aging buckets, when measured against the needs of today’s organizations, can offer a limited view of performance, often capturing symptoms rather than causes.

“Those are great early indicators, but they’re not truly representative of the illness,” Ruda said.

“If I already know my days delinquency, now it becomes a question of the next level of causation,” he added, highlighting that the result can be a shift toward more bespoke, business-specific KPIs that tie collections performance directly to broader operational outcomes.

AI changes the equation by making data more accessible and analysis more immediate. Reports that once required weeks of manual effort can now be generated in real time, freeing teams to explore deeper questions around causality and context.

“AI for AR is just taking unstructured information and turning it into outcome,” Ruda said, adding that over time, that capability can expand, enabling AI to handle a broader range of scenarios without human intervention.

From Back Office to Strategic Engine

Looking ahead, the trajectory points toward increasing autonomy, but not total automation. As models become more sophisticated, they will take on a greater share of routine tasks, particularly those involving structured decision-making. But for all its promise, AI is not a panacea.

One of the more persistent misconceptions, Ruda argued, is that automation will eliminate the need for human involvement altogether.

“The idea that AR is getting automated … can be a little scary,” he acknowledged. But in practice, AI remains heavily dependent on human input. “The AI is only as good as the one giving it the context.

“There’s always going to be those customers that just need a human to step in,” Ruda added.

In that sense, the future of collections is not just about getting paid faster but about reimagining how work gets done.

Or, as Ruda put it, it’s about enabling teams to “extrapolate your value” and operate at a scale that was previously unimaginable.

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.