Here’s the paradox facing most mid-market finance teams right now: you have more AR tools than ever, and you’re still spending half your week on manual work.
That’s not a technology gap — it’s an integration gap. And it’s exactly what Flywire set out to understand when they commissioned a survey of 300+ U.S. finance professionals earlier this year. The results, published in Flywire’s Beyond Automation: What Finance Teams Really Need from AI report, paint a clear picture of where AR stands in 2026, why automation alone isn’t enough, and what finance leaders are actually hoping AI will do for them.
Here’s what stood out.
The real bottleneck isn’t effort — it’s fragmentation
Every respondent in the Flywire survey said they deal with manual work that slows down getting paid. But the root cause isn’t that teams aren’t working hard. It’s that the systems they’re working in don’t talk to each other.
More than half of companies using enterprise software like NetSuite, Xero, or QuickBooks are bolting on as many as four additional tools to manage AR — a payment processor, a CRM, an AR automation tool, spreadsheets. Each one is solving a piece of the problem. None of them are solving it together.
The result: 83% of respondents cite poor integration between AR processes and systems of record as their biggest challenge. Another 83% say they lack visibility across their AR systems. And 70% say they have access to data — but can’t draw any real insight from it.
The most-cited day-to-day bottlenecks tell the same story: 26% say data entry between systems is what slows them down most, 26% point to following up on overdue invoices, and 25% to cash application and reconciliation. These aren’t hard problems. They’re repetitive ones — exactly the kind AI should be handling.
Getting paid is harder than it should be
Beyond the internal fragmentation, the Flywire research surfaces something finance teams often feel but rarely quantify: customers find it too hard to pay.
A third of respondents (33%) say the biggest bottleneck in getting paid faster is payment friction — customers don’t have enough options, or the process is too cumbersome to complete quickly. Meanwhile, 42% of teams are still relying on manual review — gut feel based on customer relationships and history — to figure out how to handle overdue payments.
What respondents say they actually need: smarter automation that spans the entire AR lifecycle. Automated payment matching. Deeper root cause analysis on recurring late payments. Prioritization of overdue invoices based on risk and value. AI that reasons about customer history and adapts — not just follows preset rules.
The manual workarounds are a symptom. The underlying need is for systems that make both sides of the transaction easier.
The budget is there. The confidence isn’t (yet).
One of the more striking findings in the Flywire report: 90% of those surveyed have already allocated budget for AI-enabled tools, and two thirds are actively exploring solutions. The demand is real.
But so is the uncertainty. Nearly half of respondents say they’re only “somewhat familiar” with what AI can actually do for functions like payment matching and dunning. A full third say the ROI isn’t clear and they need guidance on where to start. And the concerns are specific:
-
- 39% cite lack of trust or transparency as their top worry
-
- 37% are concerned about accuracy
-
- 35% say integration with existing systems is too hard
-
- 34% flag data privacy as a concern
None of these are reasons to slow down. They’re requirements for what good AI implementation looks like.
What good AI actually looks like
The Flywire report translates those concerns into a clear checklist for evaluating any AI solution. It’s worth keeping this front of mind as you assess your options.
AI should act, not just analyze. Finance teams don’t need another dashboard. They need AI that sends reminders, matches payments, and resolves exceptions — with humans approving the high-stakes decisions.
Embedded AI beats bolt-on tools. Given that 83% say poor integration is their core problem, AI layered on top of disconnected systems just adds complexity. The right solution is AI that’s native to your AR platform and ERP.
Human oversight is non-negotiable. With 39% citing trust as their top concern, the answer isn’t pulling back on AI — it’s building in guardrails. Approval workflows, audit trails, and clear limits on autonomous action build the confidence teams need to actually use the technology.
Your data should power your workflows — not someone else’s AI. A vendor’s AI model should never be trained on your financial data. The context that makes AI useful (customer payment history, contract terms, relationship nuance) should stay yours.
Start with high-volume, low-risk workflows. Payment matching. Reminder sequences. Data extraction. These are the right first moves — prove value on repeatable tasks before expanding to complex decisions.
What finance leaders are asking AI to do
The Flywire report asked respondents to rate how valuable specific AI capabilities would be for their work. The results show remarkable consensus on what the AR automation wish list looks like in 2026:
-
- 98% want automated payment matching
-
- 97% want auto-configuring workflows and integrations
-
- 96% want real-time cash flow forecasting and automated payment plan approvals
-
- 95% want AI to recommend which accounts to prioritize, draft collection messages, and extract data from documents
-
- 94% want AI that predicts which customers will pay
That’s not a wish list — it’s a roadmap. The technology to do most of this exists today. The gap is in finding platforms where it’s embedded, integrated, and built with the guardrails that give finance teams actual confidence in using it.
The trajectory is clear
Despite the concerns, finance leaders are bullish on where this goes. Sixty-four percent say AI will be important or essential to staying competitive in the next two years. And 99% — nearly every respondent — believe that within five years, some or all of finance will rely on AI.
The teams that pull ahead won’t be the ones who waited for perfect clarity before acting. They’ll be the ones who started with the right workflows, chose embedded over bolt-on, and built the organizational muscle for exception handling and AI supervision now.
The Flywire Finance AI Survey was commissioned by Flywire and conducted by Regina Corso Consulting in December 2025, among 300 U.S. finance professionals with responsibility for AR, cash flow, and collections at companies with $10M–$500M in annual revenue across software, SaaS, business services, marketing, media, insurance, and legal industries.
Want to see the full data? Flywire’s complete Beyond Automation report includes the full survey findings, peer benchmarks, and a practical roadmap for implementing AI in your AR process.
