Manufacturing

Toshiba reduces procurement file processing time by 82% with fileAI APIs

Toshiba reduces procurement file processing time by 82% with fileAI APIs

Toshiba logo
Toshiba logo

82% reduction

82% reduction

File processing time

20x increase

20x increase

File handling capacity

>60% OPEX

>60% OPEX

Operational cost savings

Index

Get started

Bring us your messiest documents.

See how fileAI reads, validates, and routes them.

Get started

Executive summary

Toshiba, a leading multinational electronics manufacturer, uses fileAI to address inefficiencies in processing procurement files. Previously, manual handling averaged 92 minutes per file. By embedding fileAI APIs, Toshiba automated the workflow for multilingual (English, Mandarin, Thai, etc), printed, and handwritten invoices, packing lists, and bills of lading. As a result, file processing time was reduced by 82% to just 17 minutes. This substantial improvement allowed Toshiba to significantly cut operational costs, reallocate personnel to high-priority tasks, and scale document processing capacity by 20x.

Customer profile

Industry: Electronics Manufacturing
Revenue: US$24.8B (2023)
Employees: 95,109
File Types: Invoices, packing lists, bills of lading
Languages: multilingual

Technicians in cleanroom suits working at microscope stations on an electronics manufacturing line

Business challenge

Toshiba faced significant inefficiencies with manual processing of procurement files. Each file required extensive manual entry, validation, and processing, leading to slow turnaround times, high labor costs, and limitations in operational scalability.

Objectives

  • Reduce the average file processing time significantly.

  • Improve accuracy and reliability in file handling.

  • Scale processing capacity to meet increased operational demands.

  • Reallocate manpower to higher-value business activities.

Solution architecture

Solution architecture: Accounts Payable documents (purchase orders, invoices, delivery orders, receipts) and Accounts Receivable data (sales invoices, POS data, online sales data, payment gateway) flow into fileAI via AI file capture and data extraction. fileAI file processing runs AI verification, item detail categorization, tax code and entity matching, auto 3-way matching, verification, fraud checks and payment final approval, exchanging data with external sources (vendors, customers, agents) under smart approvals based on business requirements. Outputs go to payment (disbursal of funds), data push to ERP and bookkeeping systems, and report generation, analytics, data lakes, BI/MIS tools and downstream systems, or trigger a new AI workflow.

API endpoints and middleware components

fileAI API endpoint includes:

Role

fileAI AI OCR Extraction

Extract multilingual information from documents

Classification

Zero-shot classification without upfront training

Validation

Validation of data with AI

Secure REST API Gateway

Ensure secure transmission and compliance

Implementation stats

  • Phases: Testing, Implementation, Go-live

  • Duration: 6 weeks

  • Key activities:

    • Integration of fileAI API

    • Testing and validation

    • Training and change management

Key features leveraged

  • fileAI AI OCR

  • fileAI AI Classification

  • fileAI AI Validation

  • Secure REST API infrastructure

Results

Metric

Before (manual)

After (fileAI)

Improvement

File processing time

>92 minutes

17 minutes

82% reduction

File handling capacity

Baseline

20x increase

1,900% volume expansion

Operational cost savings

Baseline

Significant

>60% OPEX

Manpower reallocation

None

Extensive

Higher-priority tasks

“The implementation of fileAI reduced file processing time from 92 minutes to 17 minutes, achieving an 82% efficiency improvement. This allows users to reallocate saved time to higher-priority tasks.”

“The implementation of fileAI reduced file processing time from 92 minutes to 17 minutes, achieving an 82% efficiency improvement. This allows users to reallocate saved time to higher-priority tasks.”

Auntra Matchima
Auntra Matchima

Auntra Matchima

Import Officer, Toshiba

Key takeaways

Early engagement with stakeholders was crucial to simplify integration efforts and accelerate adoption across Toshiba’s operations. Comprehensive testing, particularly focused on multilingual and handwritten file processing, substantially reduced potential issues post-implementation. Additionally, ongoing monitoring and analysis have been key to sustaining continuous improvement and effectively adapting to evolving business requirements.

Next steps with fileAI

Looking ahead, Toshiba plans to broaden the adoption of fileAI technology to handle additional document types and business processes. Continuous optimization of the existing APIs is also underway to achieve even greater efficiency and accuracy. Moreover, Toshiba will explore deeper integration capabilities, aiming to leverage AI-driven analytics to unlock further insights and enhance procurement decision-making.

GET STARTED TODAY

Where execution creates memory

Where execution creates memory

GET STARTED TODAY

Where execution creates memory

Building the foundation for agentic AI

Building the foundation for agentic AI

Building the foundation for agentic AI