What challenges did President Information Corporation face before AI implementation?

PIC needed to update its data pipeline and rapidly deploy secure, scalable generative AI and document understanding solutions. Its retail customers needed faster staff support, food-safety workflows, ecommerce tagging, and shelf management with better data freshness.

What was the impact of the AI solution?

Data transfer time was shortened so 7,100 stores' data reaches BigQuery in about 10 minutes. Food-safety inspection preparation was reduced by 66%, allowing a retail customer to increase inspection frequency from once per quarter to once per month. Ecommerce product search and recommendations produced a 2x higher clickthrough rate. Document AI achieved 97% OCR accuracy with about 2 seconds inference time, supporting higher-efficiency shelf management.

Source

GCP

PIC (Taiwan) deploys Gemini, Vertex AI, and Document AI for retail staff Q&A, food-safety summarization, and shelf management

Use case typeRetail analytics platformUpdated Jun 13, 2026

President Information Corporation (PIC), an IT solution provider in Taiwan and a Uni-President Group shareholder, built Google Cloud–based AI solutions to improve convenience-store operations across retail, logistics, ecommerce, and food safety. PIC streams store data from more than 7,100 convenience stores through Pub/Sub into BigQuery in about 10 minutes, enabling near real-time sales and stock visibility for operational decisions. The company also deployed a Google Cloud AI portal on Google Kubernetes Engine so enterprise customers can access customizable AI services in one place.

Organization: President Information Corporation
Industry: Retail
Location: Taiwan
Published: January 2024

Reported outcomes

quantified impactOther quantified impact

10 minutestime−66%quantified impact+97%time2 secondstime

Strategic outcomes

New product / capabilityBuilt AI tools for retail operationsBetter decisions & insightEnabled near real-time operational visibilityCustomer experience & trustImproved food-safety inspection workflowsCompetitive differentiationImproved ecommerce search relevance

Primary read

Use case focus

Showing 3 of 6

1Generative AI for staff support
2Food safety workflow automation
3Retail ecommerce search and recommendation

PIC needed to update its data pipeline and rapidly deploy secure, scalable generative AI and document understanding solutions.
Its retail customers needed faster staff support, food-safety workflows, ecommerce tagging, and shelf management with better data freshness.

PIC uses BigQuery and Pub/Sub to move operational data from edge systems into a real-time warehouse.
Using Gemini on Vertex AI, PIC developed an education chatbot for staff Q&A, a financial chatbot for daily executive reporting, and a food-safety summarization system for daily news.
PIC uses Gemini to tag product categories for vector search, improving ecommerce search relevance and recommendations.
For smart shelf management, PIC uses Document AI to extract product names and expiration dates from labels.
PIC also uses Vertex AI Vision for in-store purchase behavior analysis and Google Kubernetes Engine to host its AI portal.

Technologies

BigQuery Pub/Sub Vertex AI Gemini Document AI Vertex AI Vision Google Kubernetes Engine

Data transfer time was shortened so 7,100 stores' data reaches BigQuery in about 10 minutes.
Food-safety inspection preparation was reduced by 66%, allowing a retail customer to increase inspection frequency from once per quarter to once per month.
Ecommerce product search and recommendations produced a 2x higher clickthrough rate.
Document AI achieved 97% OCR accuracy with about 2 seconds inference time, supporting higher-efficiency shelf management.

Architecture

Store data is streamed through Pub/Sub into BigQuery. PIC builds generative AI applications with Gemini on Vertex AI, uses Document AI for label extraction, Vertex AI Vision for store-behavior analysis, and deploys a customer-facing AI portal on Google Kubernetes Engine. The article also mentions a third-party vector database for retrieval and tagging workflows.

Sources & evidence1

Groundedness: 5/5

Primary source

AI-generated summary. Verify important details with the linked sources before relying on this case.

Explore related AI use cases

Customer evidence

Provider evidence

GCP provider page GCP ranked cases

Market context

Retail AI use cases Retail industry insights Taiwan country insights

AI patterns

Vision