AI Hub
News, Reports, and Cutting-Edge Case Studies
What’s New
- Nov. 22, 2024 Report Using AI for Fraud Detection
- Nov. 22, 2024 Report The Power of Generative AI in IT Departments
- Sep. 10, 2024 News Release NRI to Unveil Leading-Edge Financial AI Platform in 2025 Designed for Optimal Data Security and Data Sovereignty
- Sep. 10, 2024 Publications Advantage of Japanese approach to AI regulation
- Aug. 13, 2024 Publications Promoting generative AI utilization in the financial sector
Case Studies
ALL
In helpdesk operations, the experience level of staff greatly impacts efficiency. It is difficult to find the necessary information from the vast amount of manuals and documents accumulated within the company. Less experienced members often spend significant time and effort finding the correct answers, leading to disparities in response capabilities.
A chatbot is trained to learn documents and business characteristics of the user company. Utilizing generative AI like ChatGPT can create a user-friendly chatbot interface. This system is accessible even for less experienced personnel, streamlining the process of finding necessary information and improving the efficiency and consistency of responses to inquiries.
- Quick and accurate answers for less experienced team members.
- Reduced disparities in response capabilities among staff.
- Enhanced knowledge sharing across the organization, fostering continuous improvement and operational excellence.
ALL
Customers struggle to analyze vast amounts of purchase information, complaints, encouragement calls, and social media feedback. Existing tools often fail to meet their needs, and building new systems is costly and time-consuming.
- NRI helps customers visualize and organize their challenges.
- NRI leverages over 20 years of its research and development expertise in Japanese text analysis.
- NRI offers modular analysis functions that can be easily integrated into existing systems.
- Automatic classification and analysis of call center inbound content
- Risk management through the collection and classification of SNS data
ALL
Preparing FAQs, transcribing customer conversations, and analyzing transcripts require significant time, resources, and training. Inexperienced helpdesk members struggle to transcribe customer conversations and quickly extract and record important information
It enables the quick extraction and recording of important information without individual differences among members. The data can be used for analyzing the Voice of the Customer (VoC) and creating FAQs.
- Reduced response time to questions.
- Accelerated transcription and summarization of customer conversations.
- Enhanced quality of Voice of the Customer analysis.
Banks (Yamaguchi Bank, Momiji Bank, Kitakyushu Bank)
Utilization of accumulated negotiation histories and solution proposals has been limited to individual sales representatives.
AI will recommend solutions and past examples from the negotiation history data of all employees, based on similar negotiation records from previous sales activities.
Enhancing service quality and customer satisfaction by enabling all employees to propose optimal solutions to customer challenges.
Insurance Company
Processing insurance claims required manual entry and verification of hospitalization or surgery certificates, which was labor-intensive and error-prone due to high volumes.
- Selected AI-OCR products balancing accuracy, speed, and cost.
- Implemented measures to enhance reading accuracy, including automatic form correction and strategies for recognizing handwritten medical terms.
- Designed a system integrating human oversight to maintain accuracy and consistency.
- Achieved over 94% AI-OCR recognition accuracy for forms.
- Enabled processing of 7,000 to 10,000 forms per day, converting them into digital data.
- Streamlined workflow by replacing some manual data entry tasks with AI-OCR processing.
Retail
Regular visitor counts and tenant sales analyses are conducted, but real-time analysis of customer demographics and behavior across the entire mall is not yet achieved, limiting marketing efforts.
- Install cameras in multiple store locations for image processing and monitoring.
- Use image analysis to track real-time visitor information, including gender, age, number, and movement paths, enabling strategic placement of digital signage advertisements.
- Effectively guide customers to stores and optimize product assortment.
- Efficiently attract target customer demographics in tenant stores, potentially improving profitability.
- Provide detailed and personalized guidance by detecting VIP customers and those with strollers.
Retail
Japan faces a significant labor shortage with the working-age population decreasing by nearly 500,000 people annually.
- Shoppers order items via a smartphone app, and in-store robots pick and place products in a collection box. After payment, shoppers retrieve their items from the collection box.
- Only during times when staff cannot be available, such as at night, the store transforms into a large vending machine.
- Extend business hours at minimal cost, even for stores that close at night.
- Unlike other services (such as Amazon Go), there's no need for numerous cameras and sensors on the ceiling or shelves; minimal renovations are sufficient. Achieve unmanned operations with minimal renovations by using robots for product retrieval and delivery
Retail
With a declining workforce due to an aging population, the retail industry faces increasing pressure for meticulous management, analysis, and merchandising. Changing the layout of sales floors requires significant effort to register product positions in the management system, risking errors.
Attach ArUco markers with embedded shelf position information to shelves. Use a smartphone or tablet camera with a reading app to scan product barcodes placed between the markers.
- Determine product locations on shelves based on scanned ArUco markers and barcodes.
- Manage orders and merchandising directly on the sales floor by displaying sales and inventory information, reducing the need to return to the back office and streamlining workflow.
A significant concern with utilizing generative AI is the "risk of data leakage." For customer companies using external generative AI services, determining how much confidential information can be safely transmitted remains a critical issue.
NRI's "Private LLM" secures sensitive information with private cloud or on-premises deployment and can be tailored to business needs. It also includes versatile modules like "Private Speech Recognition," expanding its application across various industries.
Companies with high information security demands can safely utilize generative AI. Tailored AI customization for specific business needs will enhance operational efficiency and drive the creation of new value.
Inquiries
For any inquiries, please email pr_inquiries@nri.co.jp.