Shortens business data transfer time to 10 minutes with Pub/Sub
Enhances food-safety inspection efficiency by 66% with Gemini
Increases clickthrough rate by 2X with more relevant product search results and recommendations
Realizes 97% optical character recognition accuracy rate to enable smart shelf management
By leveraging Google Cloud to develop and run its AI solutions, PIC has quickly built an AI portal offering diverse, customizable, highly available AI-powered services that help enterprises across industries optimize their operations.
By supporting low-latency connectivity in Taiwan, providing industry-leading data processing and AI capabilities and enabling seamless integration with open-source tools, Google Cloud can help us develop innovative AI solutions faster and more effortlessly.
Tony Hsieh
General Manager, PIC
As the benefits of artificial intelligence (AI) adoption become increasingly prominent, many companies are actively looking for ways to leverage the latest AI technologies easily and effectively.
President Information Corporation (PIC) is dedicated to making state-of-the-art AI solutions customizable and accessible for enterprises across industries. Founded in 1997, PIC is an IT solution provider specialized in retail business, logistics, enterprise resource planning, and supply chains. With Uni-President Group — one of the largest food manufacturing and retail operators in Taiwan — as its largest shareholder, PIC has accumulated profound expertise in retail-store management by enabling the digitization of more than 7,100 convenience stores over the past 20 years.
"Part of our competitive edge comes from our decade-long experience of supporting convenience stores' 24-hour operations," notes Tony Hsieh, General Manager at PIC. "We're specialized in providing reliable IT solutions that empower high operational efficiency and cover all aspects of retail-store operations from logistics to food safety to fraud prevention."
To provide better IT services, PIC has always been at the forefront of adopting new technologies. In 2019, the IT solution provider founded a data department with 30 employees to use big data in its solutions and implement data-driven strategies. As AI matured in 2023, PIC decided to explore potential use cases for the emerging technology by establishing an AI team. However, the team soon realized that to drive up AI innovation, it needed to update its data pipeline and acquire powerful AI model training tools. With a data center in Taiwan that supports faster data transfer, Google Cloud was the obvious choice of platform for PIC to use to develop and run its AI solutions.
"To take full advantage of AI, having a high-performance, highly secure data warehouse is a crucial prerequisite," explains Hsieh. "By supporting low-latency connectivity in Taiwan, providing industry-leading data processing and AI capabilities and enabling seamless integration with open-source tools, Google Cloud can help us develop innovative AI solutions faster and more effortlessly."
As the first step of its migration to Google Cloud, PIC started using BigQuery as its enterprise customers' data warehouse in Taiwan. With Pub/Sub streaming business data from end devices to BigQuery and the low latency enabled by the local Google Cloud data center, PIC can shorten data transfer time and realize real-time databases for better decision-making.
For example, PIC deployed the data pipeline of its convenience store customer in Google Cloud, the 7,100 store data is sent to BigQuery through Pub/Sub in 10 minutes. This way, the convenience store chain is informed about the current sales and stock status almost instantly. It can, therefore, adjust its marketing and logistics plans more promptly to eliminate unnecessary costs.
Our enterprise customers can process more than 60 gigabytes of data every day, so having high-performance data tools like BigQuery and Pub/Sub can help us effectively optimize our customers' operations.
Areoll Wu
Vice General Manager, PIC
"Our enterprise customers can process more than 60 gigabytes of data every day, so having high-performance data tools like BigQuery and Pub/Sub can help us effectively optimize our customers' operations," explains Areoll Wu, Vice General Manager at PIC. "By turning data processing from in batch to streaming, we empower our customers to enhance the cost-effectiveness of their data-driven strategies."
Gemini has helped us greatly improve our customers' operations and services not only with its industry-leading generative AI capabilities, but also through the diversity of use scenarios where it can come in handy.
Areoll Wu
Vice General Manager, PIC
Having established data pipelines centered around BigQuery and a third-party vector database, enabling efficient data retrieval, PIC started training AI models that benefit its customers' business operations. Leveraging Gemini with Vertex AI, PIC developed several generative AI solutions, including an education chatbot that allows retail store staff to easily find detailed responses to their questions, a financial chatbot that generates daily financial reports and insights for executives, and a food-safety system that automatically summarizes related daily news in five to 10 minutes.
According to Wu, the automation enabled by Gemini has brought significant benefits. For example, since complex preparation work was required, a retail customer of PIC could only conduct a food safety inspection of its stores once every quarter.
With the AI features of PIC's food safety system, the preparation work can now be completed 66% faster, and the retail company can inspect its stores once every month, which further guarantees its product quality.
On top of that, to improve the online shopping experience provided by its ecommerce customers, PIC uses Gemini to identify product categories and tag product data for vector search. As the data is highly accurately tagged, PIC's ecommerce product search and recommendation system can provide more relevant search results and product recommendations, which results in a 2X higher clickthrough rate on its customers’ sites.
"Gemini has helped us greatly improve our customers' operations and services not only with its industry-leading generative AI capabilities, but also through the diversity of use scenarios where it can come in handy," says Wu. "It enables a wide range of our AI solutions from staff training to fraud detection, food safety, and ecommerce."
Besides Gemini, PIC also leverages other pre-trained AI models in Vertex AI to develop its AI solution. To enable smart shelf management for retail stores, the PIC team uses Document AI to extract information like product names and expiration dates from product labels. As Document AI supports 97% optical character recognition (OCR) accuracy rate, with an inference time of two seconds, PIC's shelf management system can quickly identify products displayed on shelves and support higher management efficiency.
To realize omnichannel marketing, the PIC team uses Vertex AI Vision to identify customer purchase behavior at physical stores and matches the data with the online user behavior data gathered through Google Analytics. Its product recommendation system can then produce more relevant results by analyzing users' preferences both online and offline. Furthermore, the team has trained a smart vending machine AI model that supports product identification and cashierless checkout using Multimodal AI.
Whenever we want to develop an AI feature, 80% of the time we can find a pre-trained model available in Vertex AI. We wouldn't have been able to release so many AI solutions within a year if there were no resources provided by Vertex AI.
Areoll Wu
Vice General Manager, PIC
"Vertex AI has facilitated our AI development work to a great extent. Whenever we want to develop an AI feature, 80% of the time we can find a pre-trained model available in Vertex AI," notes Wu. "We wouldn't have been able to release so many AI solutions within a year if there were no resources provided by Vertex AI."
To streamline AI adoption for its customers, PIC launched an AI portal deployed in Google Kubernetes Engine (GKE) that gathers all its customizable AI solutions in one place. By inputting their own business data, PIC's enterprise customers can easily access various AI features that perfectly answer their requests. With the high scalability enabled by GKE, PIC's AI portal can provide reliable services under different circumstances, fulfilling business needs across industries.
Google Cloud has enabled us to reach and even surpass our AI innovation and development targets with its outstanding AI and data processing solutions. We're now ready to help companies in different industries catch up with the trend and benefit from AI.
Tony Hsieh
General Manager, PIC
Following the successful launch of its AI portal in Taiwan, PIC is expanding its business into more industries and foreign markets. Hsieh says that while discussing with potential customers in Southeast Asia, Translation AI has helped enhance communication efficiency by providing accurate translations between multiple languages. At the same time, PIC will continue optimizing its AI solutions by adding new innovative features, such as stock-management suggestions generated by an AI model based on sales prediction. It also plans to make its AI solutions available on Google Cloud Marketplace to reach more customers.
"Google Cloud has enabled us to reach and even surpass our AI innovation and development targets with its outstanding AI and data processing solutions," notes Hsieh. "We're now ready to help companies in different industries catch up with the trend and benefit from AI."
Founded in 1997, President Information Corporation (PIC) is an IT solution provider specialized in retail business, logistics, ERP, and supply chains. Uni-President Group, PIC's largest shareholder, is one of the largest food manufacturing and retail operators in Taiwan. Committed to providing outstanding service to its enterprise customers, including 7-ELEVEN stores, PIC established an AI team in 2023 to develop innovative AI-powered features.
Industry: Retail
Location: Taiwan
Products: Vertex AI, BigQuery, Cloud Storage, Document AI, Google Kubernetes Engine, Multimodal AI, Pub/Sub, Vertex AI Vision