This case study illustrates SHIFT ASIA’s ability to create innovative, secure, and scalable AI-powered solutions tailored to our clients’ specific requirements, improving operational efficiency and customer satisfaction.

About the Client

The Client is a leading IT services provider in Japan. The company offers a wide range of services, from system development and data analysis to cloud services. It is well-known for its cutting-edge technology, such as AI integration, machine learning, and cloud technologies. With expertise covering finance, healthcare, manufacturing, and the public sectors, the company delivers high-quality solutions and IT strategies to help businesses optimize their operations, improve business efficiency, and achieve sustainable growth.

Challenges & issues

The Client came to SHIFT ASIA with two main projects that they were facing development obstacles:

• eCommerce GPT project: initially built for their eCommerce system to generate products SEO content. The challenge was to create unique and engaging content for each product, avoiding repeating content.
• Travel GPT project: developed for a travel agency, creating content based on public information from their website and proprietary data, including confidential PDF files. The system was to generate original content while handling this sensitive data with utmost care.

SHIFT ASIA’s approach

We realized that both projects required processing a huge amount of data and information and the complexity of creating multi-level content. Hence, we proposed a development plan focused on security, efficiency, and scalability.

eCommerce GPT:

• Built a proprietary system on Azure to ensure security
• Integrated with the Client’s existing database, utilized Azure search function for content analysis, and delivered correct multilingual content.

Travel GPT:

Utilizing Azure services, we built a solution to

• Analyzed and interpreted public data
• Retrieved information from the Client’s website correctly via our proprietary prompts.
• Developed custom ChatGPT models with K-means algorithms for data clustering, SQL query conversion, and providing user responses.
• Test data mining using confidential PDF files.

Results

Through regular communication with the Client throughout the project, we received instant feedback, performed continuous testing, and refined the system. Both solutions have been successfully integrated into the Client’s production environment. As a result, the project was praised and extended for excellent work.