Webinar: How to Free the $600B Trapped in Accounts Receivables

Building an AI strategy to generate cash faster

Businesses have already earned the money sitting in accounts receivable. The problem is getting that cash from the balance sheet to the bank account faster.

Watch the webinar replay to hear the full conversation and see how AR teams can move from AI interest to an executable plan.

Cash Generation starts with finding out where cash is trapped

The webinar begins with a simple idea: trapped cash is revenue a business has earned but cannot yet use.

That cash can get stuck almost anywhere in the AR process. An invoice may go out late or with incorrect information. A dispute can move between departments without a clear owner. A collector may spend time on the wrong accounts. A buyer may be ready to pay but face unnecessary payment friction. Or the money may already be in the bank while the AR team is still trying to match it to the correct invoices.

“Your organization has earned that revenue. You’ve delivered the product or the service. You’ve sent the invoice, but the cash still hasn’t reached your bank account.”

Mark Brousseau, President of Brousseau & Associates

The audience poll reinforced how widely these problems can appear. Among respondents, 42% identified deductions or dispute resolution as a likely place for cash to become trapped, while another 42% selected customer payment processes. Collections prioritization and follow-ups were at 37%. The point is not that every company has the same bottleneck. It is often the case that the delay begins before an invoice becomes past due. For finance leaders, that changes the question. Instead of asking only how to collect faster, look across the credit-to-cash cycle and identify where work, decisions, or missing information are slowing the movement of cash.

Choose AI use cases by their Cash Generation potential

One of the webinar’s clearest recommendations is to start with the business problem, not the technology. Brousseau encourages AR teams to look for work that is frequent, variable, and dependent on human interpretation. These are often stronger AI candidates than tasks already handled well by fixed rules. Potential opportunities can appear throughout AR. AI may help identify invoice errors, detect delivery problems, prioritize collections activity, classify disputes, recognize payment behavior, or improve the matching of payments and remittance information. But an interesting use case is not automatically a valuable one.

The webinar recommends assessing opportunities across several practical criteria:

• Volume: Does the issue occur frequently enough to matter?
• Variability: Do changing formats and scenarios make fixed rules difficult?
• Data: Can the team connect usable inputs with past decisions and outcomes?
• Cash impact: Does the problem delay cash or make the cash position harder to understand?
• Exceptions: Does the work consume experienced employees’ time because judgment is required?

Teams can then rank possible use cases based on cash impact, feasibility, manual effort, control requirements, and time to measurable results. That discipline helps prevent a common failure: adopting AI because leadership wants an AI initiative rather than because the business has identified a problem worth solving.

Weak projects often begin when “the goal was to just use AI and check the box.” A better question is: What specific delay or decision are we changing, and how will that change affect cash?

Cash Generation depends on more than clean data

Data quality matters, but it is only one part of AI readiness. In the webinar’s second poll, 58% of respondents identified integration with existing systems and workflows as their biggest obstacle to deploying AI in AR production. Another 36% selected incomplete, fragmented, or unreliable data. Those results point to a broader readiness test. First, look at the process. Is the workflow stable enough to improve, or would automation simply preserve unnecessary steps?

Then consider the data. Can the organization trace where information came from, how it was used, and what happened as a result? Ownership matters too. A meaningful AI initiative needs a business owner who can define the outcome, make workflow changes, monitor performance, and decide whether to scale, revise, or stop the project. Controls also need to be clear. Finance teams should be able to review key recommendations, understand the information behind them, and determine where human approval is required. Finally, AI has to fit into daily work. A technically capable tool that creates another disconnected process can add complexity rather than remove it.

Treat the pilot as a business test

A successful AI pilot should answer a defined business question.

• Can better collection prioritization lead to earlier action on accounts where intervention can change payment timing?
• Can AI improve automatic payment matching without creating unacceptable posting errors?
• Can better dispute routing reduce the time cash remains blocked?

Those are useful questions because the outcomes can be measured. Before the pilot begins, Brousseau recommends defining the scope, baseline, guardrails, success measures, and criteria for moving forward.

“Please don’t let a vendor tell you what success looks like. Make sure you know what success looks like.”

Mark Brousseau, President of Brousseau & Associates

That means establishing the starting point before introducing new technology. If a team does not know its current cycle time, exception rate, match rate, workload, or other relevant measures, it will have difficulty proving improvement later.

Human oversight should also be designed into the pilot based on risk. Low-risk, high-confidence activity may require less intervention. High-consequence decisions deserve more review.

The objective is not for employees to recheck every AI-assisted action. It is to put human judgment where it makes the biggest difference. Frontline employees should be part of that process. They know where exceptions occur, which workarounds keep the process moving, and where a new system could create problems. Their involvement helps determine whether the workflow will work outside a controlled test.

Scale only when the whole workflow works

Technical accuracy alone does not make an AI pilot ready for production.

Mark identifies five gates that should guide the decision to scale:

1. Technical accuracy
2. User adoption
3. Control performance
4. Operational capacity
5. Measurable business impact

A weakness in one area can undermine strengths elsewhere.

If the output is accurate but employees do not use it, the initiative has not changed the process. If users adopt it, but exception handling creates more work, capacity has not improved. If the workflow performs well but the organization cannot connect the change to a meaningful business result, the case for expansion remains weak.

This is also where clear accountability matters.

“If everyone owns AI, no one owns the outcome.”

Mark Brousseau, President of Brousseau & Associates

Finance, IT, risk teams, frontline users, and technology providers may all contribute. But one business owner needs responsibility for the result.

Connect AI activity to Cash Generation

For the CFO, model performance is only part of the story.

The stronger case connects four stages: AI output → changed action → operational result → financial effect

For example, better prioritization of collections may lead to earlier action. Earlier action may improve resolution or payment timing. Those changes can then be evaluated against overdue balances, DSO, working capital, or cash forecasting.

This is why “reduce DSO” is too broad to serve as an AI use case on its own. DSO is influenced by payment terms, sales mix, seasonality, and other factors. Instead, identify the operating lever the initiative can actually change. For invoicing, that may be billing cycle time or delivery success. For collections, it could be time to action. For disputes, track dispute age. In cash application, match rates and unapplied cash aging provide clearer leading indicators. Cash predictability matters as well. Payment behavior, disputes, broken promises, and other signals can help finance teams form a better view of when cash is likely to arrive. That does not make a forecast certain.

A probability with transparent assumptions is not a promise, but it is really valuable. Better visibility gives CFOs more information for liquidity, working capital, and funding decisions.

Build a 90-day path to Cash Generation

The webinar closes with a practical three-step approach.

Days 1–30: Map where cash slows down, establish baselines, and rank possible AI use cases.

Days 31–60: Select one meaningful pilot, set controls and review thresholds, and prepare the data and workflow.

Days 61–90: Run the pilot, compare performance with the baseline, document what worked, and decide whether to scale, refine, or stop.

The timing can vary. The sequence should not. AI creates value in AR when it changes a real workflow, employees trust and use it, controls hold, and the results connect to cash. That is the larger lesson from the webinar. Start with the cash, not the technology. Find where earned revenue is getting stuck. Measure the problem. Test a specific improvement. Then follow the evidence from the work all the way to the financial outcome.

Billtrust is the B2B Cash Generation Platform, bringing invoicing, payments, credit, collections, and cash application together around a common goal: converting earned revenue into cash you can predict, defend, and deploy.

Learn more about Billtrust and how Cash Generation can help your finance team release trapped cash across the credit-to-cash cycle.