AI-Driven Open Item Matching

The Challenge

A global FMCG enterprise was processing thousands of open items daily through S/4HANA's F.13 clearing run, with reconciliation handled manually by finance teams across multiple regions. The challenge was designing an event-driven architecture that could capture open items in real time as F.13 detected them, feed them into an AI matching engine without polling or batch delays, and present validated recommendations in a user-facing cockpit, without disrupting the existing S/4HANA landscape or requiring changes to core financial processes.

Technology

S/4

S/4HANA

EM

Event Mesh

CAP Service

SAP AI Core

HANA Cloud

Architecture

My Role

• Led solution architecture from discovery through design, translating finance team requirements into a BTP-native event-driven architecture
• Designed the Event Mesh integration layer to capture S/4HANA F.13 open items in real time as they are detected
• Architected the CAP application layer for event ingestion, AI orchestration, user validation, and action completion
• Defined the SAP AI Core matching model integration and the confidence-based recommendation framework for finance users
• Delivered architecture blueprints, design documentation, and stakeholder workshops throughout the engagement

Outcomes

• Architecture design approved and handed to development team for implementation
• Event-driven pattern eliminates batch processing delays: open items available for AI matching in real time
• AI matching layer designed to handle thousands of items per day without manual intervention
• Solution designed as a reusable pattern applicable to other S/4HANA finance clearing processes