Yooz launches agentic AI for non-PO invoice coding

Invoices without purchase orders often require finance teams to manually assign accounting or analytical codes at the line-item level. This task can be repetitive and time-consuming, particularly as invoice volumes rise or for recurring invoices needing consistent coding. Yooz's Agentic AI Allocation is designed to address this by applying or suggesting these allocations, aiming to streamline a traditionally hands-on part of financial operations.

The software operates on a confidence-based system: it automatically fills in allocations when its confidence is high, offers recommendations for review when confidence is moderate, and leaves coding entirely to a user when confidence is low. The system reportedly improves its recommendations through completed and validated workflows, with a tool called Yoozie flagging AI-applied actions for user visibility. Yooz CEO Laurent Charpentier stated that AI creates the most value when built directly into daily financial processes.

This launch extends Yooz's existing line-level automation tools, which previously focused on invoices linked to purchase orders. The company states it reinvests 25% of its income into research and development each year, contributing to over 100 scientific publications in artificial intelligence and machine learning. Senior Product Marketing Manager Alastair Moore noted the goal is to bring intelligence directly into line-level coding decisions, allowing finance teams to automate more of the process while maintaining visibility and control.

The development aligns with a broader trend in accounts payable software, where competition increasingly centers on embedding artificial intelligence directly into day-to-day processing. This includes capabilities such as document classification, coding suggestions, and exception routing within existing workflows. Yooz reports processing over 300 million financial documents and supporting integrations with more than 250 enterprise resource planning systems.