Challenge
SLA pressure and manual research were slowing everything down
The company runs a high-volume accounts payable (AP) helpdesk, fielding a constant stream of vendor emails requesting invoice or payment status updates. While the service-level agreement (SLA) required a 24-hour turnaround, responses struggled to meet the deadline, creating operational stress and leaving little room for error.
Researching the queries was the real bottleneck. Associates had to manually access SAP, validate invoice and payment data, and then construct an accurate reply. This repetitive, time-consuming process drained hours from the team and pulled it away from more high-value work.
The company recognized the data access and workflow problem, and that it needed a smarter approach to how its AP helpdesk operated.
Solution
Embedding AI directly into the AP workflow
The company partnered with Genpact to embed Genpact AP Assist, part of the Genpact AP Suite, directly into its existing workflow platform – minimizing disruption while helping drive meaningful impact.
Genpact AP Assist automatically ingests and indexes inbound vendor emails, filters and helps close spam with minimal human involvement, and identifies inquiries suited for agentic AI-assisted responses. It then retrieves invoice and payment data from SAP through a frequently refreshed Secure File Transfer Protocol (SFTP) connection to help improve accuracy and vendor trust.
It classifies each email's intent, selects the right handling path, and produces polished, comprehensible drafts that combine natural language with SAP data. Every response goes through an associate review, supporting tight governance and a human-in-the-loop process that protects quality and minimizes risk.
Impact
Faster responses, stronger relationships, and a smarter AP operation
The implementation delivered quick results. Response times dropped from as long as 24 hours to within about 8 hours. Time taken to draft responses reduced significantly, from around 30 minutes per email to just seconds. AI response accuracy climbed from about 70% to more than 90%, exceeding the stated 85% target.
But the real impact went beyond the numbers. Vendors started receiving faster, more consistent answers, contributing to fewer escalations and positive stakeholder feedback. With repetitive SAP lookups off their plates, associates are now focusing more on exceptions and higher-value work.
This engagement proved something important: embedding AI directly into core operational workflows – not just layering on another generic tool – can help drive meaningful, scalable improvements. The company has improved KPIs and redefined how its AP helpdesk operates.
