Betfred unlocks a 47% AP efficiency improvement with AI powered by Proactis

Challenge

Processing over 40,000 invoices a year across a high-volume, fast-moving retail and digital business places significant demands on any accounts payable function. For Betfred, the stakes are particularly high: invoices span a wide range of supplier types, formats and values - from large-scale operational contracts to frequent low-value transactions across hundreds of retail locations.

Prior to implementing Proactis Invoice Capture, the AP team was managing a largely manual process. Invoices arrived in multiple formats, required manual data entry into the finance system, and were subject to the delays and errors that inevitably accompany high-volume manual handling. With a team responsible for maintaining tight payment performance and strong supplier relationships across the business, the pressure to reduce manual effort - without introducing new risk - was real.

The organisation had already taken a significant step forward by deploying Proactis Purchase-to-Pay and Invoice Capture, which automated the core processing workflow and integrated directly with Unit4 Financials by Coda. But as invoice volumes remained high and the expectation of continuous improvement grew, the question became: how much further could performance be pushed?

The answer lay in AI. The specific challenge was to improve the accuracy and speed of invoice data capture at the point of ingestion - reducing the number of invoices requiring manual intervention, cutting processing time per invoice, and further reducing the risk of errors reaching the finance system. Any solution would also need to work within the existing technology stack, with no disruption to live operations during adoption.

Solution

Proactis took a proactive, partnership-driven approach - approaching Betfred directly with the opportunity to trial the AI Accelerator enhancement for Invoice Capture rather than waiting for a contract renewal or a reactive conversation.

The AI Accelerator applies machine learning to the invoice capture process, improving the accuracy with which key data fields are identified and extracted from incoming invoices. Unlike rule-based systems that rely on static templates or supplier-specific configurations, the AI model learns continuously - adapting to new invoice layouts, formats and supplier patterns over time, and improving its own performance with every document processed.

The two-week trial was designed to give Betfred a clear, evidence-based view of performance before any commitment was made. Existing live invoice volumes were processed through the AI Accelerator alongside the standard capture workflow, giving the finance team a direct comparison of accuracy rates, processing speeds and exception volumes.

The results were unambiguous: a 47% improvement in AP efficiency, with measurable gains in data accuracy and a significant reduction in the volume of invoices requiring manual review. The trial also validated the seamless integration of the AI layer with the existing Proactis and Unit4 Financials by Coda environment, confirming that adoption could proceed without any reconfiguration of the underlying platform.

Following the trial, Betfred moved to full adoption of the AI Accelerator - extending the performance gains across the entire invoice processing operation and embedding AI-driven automation as a permanent feature of the AP workflow.