Case Study: AI for AP/AR: Cutting Cycle Time While Strengthening Controls

Introduction

For a growing retail and restaurant company, accounts payable and receivable were functional, just not functioning well. Invoices moved. Payments were made. But underneath the surface, a significant share of transactions required manual intervention, and that manual work was consuming time and money the business could not afford to keep losing. When Altum Strategy Group was engaged to assess and improve the AP/AR function, what emerged was not simply a process problem. It was a data problem, a controls problem, and ultimately a financial exposure problem that AI was uniquely positioned to solve.

The Challenge

The company’s AP/AR operations ran on a combination of manual lookups, spreadsheet edits, and file uploads that bypassed the core system entirely. Approximately 20% of all invoices — both inbound and outbound — were edge cases the system could not process automatically. The causes varied: mismatched part numbers, invoice totals that did not reconcile, and supplier entries appearing under multiple names across different distribution entities. Each exception required a staff member to investigate, make a judgment call, and manually correct the record outside the system before re-uploading it.

The result was a processing backlog that had grown well beyond acceptable limits. The company’s target was a 30-day SLA. The mean processing time had climbed to 120 days. Some invoice batches had been outstanding for 180 days — meaning payments and receivables from early in the year were still unresolved by mid-year, with every subsequent month’s volume stacking on top.

The core challenges were:

• Invoice processing times were averaging 120 days against a 30-day SLA, with some batches delayed beyond 180 days
• Approximately 20% of all invoices requiring manual exception handling due to data mismatches, unresolved part numbers, and invoice discrepancies
• Underlying supplier data that was fragmented, inconsistent, and not fit for automation — the same supplier appearing under multiple entries, parent companies mapped inconsistently across subsidiaries
• Manual edits made outside the system, eliminating any audit trail of who changed what and when
• Significant staff time consumed by exception resolution, estimated at approximately $1 million annually

The Solution

Altum’s approach began not with automation, but with the foundation. Before any AI tool could be applied effectively, the underlying data had to be sound. The team invested heavily in upfront cleansing, normalization, and remapping; consolidating duplicate supplier entries, reconciling parent companies with their distribution subsidiaries, and establishing the data quality baseline that automation requires. Following the 80/20 principle, the focus was on resolving the highest-impact inconsistencies first to unlock the most value, quickly.

With a clean foundation in place, Altum deployed AI to analyze the full AP/AR data set, identify patterns across invoice types, and define the rules and thresholds that governed how different categories of exceptions should be handled. This was not guesswork. The AI surfaced the logic that had previously lived only in the heads of experienced staff — price tolerance thresholds, part number substitution patterns, supplier-specific billing conventions — and made it explicit, auditable, and automatable. That same AI engine, the one used to analyze the invoice data and surface the exception-handling logic, became the agent Altum trained the client’s team on and left behind.

Key elements of the solution included:

1. Data Cleansing and Normalization

• Comprehensive remapping of supplier records, parent companies, and distribution entities
• Data normalization prioritized by business impact, resolving the most consequential inconsistencies first to unlock automation value quickly

2. AI-Driven Process Automation

• Deployment of AI to analyze invoice data at scale, identify patterns, and define exception-handling rules
• Automation of data import, export, and manipulation, replacing the manual workflows that had created the backlog
• Rules-based processing for the most common exception types — tolerance thresholds, part number substitutions, billing convention variations — removing human bottlenecks from the highest-volume cases

3. Intelligent Agent and Capability Transfer

• The AI engine used for the initial invoice analysis was configured as the agent handed to the client’s team — not a separate tool, but the same system they had already watched work, now in their hands to run and refine
• Hands-on training and knowledge transfer so the client could adapt the system as their business evolves — not simply inherit a black box
• Shoulder-to-shoulder coaching through go-live, building the internal confidence and competency to sustain the change

4. Controls and Audit Trail

• All invoice matching and exception resolution moved into the system, creating a complete, timestamped record of every action taken
• User-level tracking replaced the uncontrolled practice of editing supplier invoices externally and re-uploading them — eliminating a material controls gap

The Results

With Altum’s implementation, the client achieved measurable improvements across both efficiency and control:

• Invoice processing cycle times reduced from a mean of 120 days to near the 30-day SLA target
• An estimated $1 million in annual staff time redirected from manual exception handling to higher-value work
• Edge case resolution automated for the majority of previously manual exceptions, freeing finance staff to focus on analysis rather than administration
• Full audit trail established within the system, replacing uncontrolled external edits with tracked, in-system actions
• Improved traceability and confidence in financial data across the AP/AR function

Conclusion

The AP/AR function is often treated as a back-office cost center — necessary, but not strategic. This engagement showed what it can become when people, process, and technology are properly aligned: a source of financial predictability, operational control, and freed capacity. For Altum, the starting point is always the same: understand the business, fix the foundation, then apply the right tools. That sequence is what separates automation that delivers lasting value from automation that disappoints.

For CFOs and finance leaders navigating similar challenges, Altum Strategy Group brings the operational depth and technical capability to transform finance operations into a competitive advantage.