SHIFT is transitioning expense settlement from human review and judgment to AI-based initial review. It is conducting a pilot aimed at having AI initially review 70%–80% of the 7,000–8,000 monthly claims, and has outlined operational challenges and future prospects.
As part of its efforts to promote the use of AI, SHIFT has established a system in which AI cross-checks and initially reviews the human judgments in expense settlement that had previously depended on the experience and intuition of skilled employees. The company has stated that it will first accumulate data and gradually introduce AI into operations without significantly changing its existing systems.
Expense settlement had already been systematised, but at the beginning and end of each month, accounting staff still had to cross-check multiple systems and verify deficiencies and discrepancies. SHIFT organised the application information and review criteria, then tested an AI-based initial review system in a pilot department.
As an operational result, the company confirmed that AI could initially review approximately 70%–80% of applicable claims. Accounting staff are transitioning to a process that prioritises claims with possible discrepancies or exceptions, while AI is also being used to organise points of non-compliance and draft notification letters to applicants when claims contain deficiencies.
The company is also confronting the realities and challenges of not being able to leave everything to AI. To place all claims in AI's hands, continued verification and sampling of review accuracy, as well as responses to unexpected applications, are essential to operational design. Rather than undertaking a large-scale system overhaul, the company is taking a gradual transformation approach that makes use of its existing systems.