Uni-President Chain Store Corporation

Uni-President Chain Store Corporation scales hyper-personalized retail with Google AI

Results on Google Cloud

PCSC unifies data silos on Google Cloud, leveraging Gemini models to deliver hyper-personalized retail to 20M app users.

Two smartphones displaying an AI food recommendation app interface with a character mascot and nutrition info

Breaking data silos and overcoming
high-concurrency tunnels
to deliver hyper-personalized retail at scale

In Taiwan, 7-ELEVEN is more than a supermarket. It is a central pillar of daily neighborhood life. Across 7,300+ stores serving nearly 20 million loyalty members, operator Uni-President Chain Store Corporation (PCSC) sits at the heart of the daily Taiwanese routines. 

To elevate this connection, PCSC set out to answer a daily question "What should I eat next?" through an intuitive generative AI assistant called the "AI Food Passport." They realized that delivering an instant, personalized answer was a monumental technical challenge.

However, a fragmented data landscape threatened the project. Transactional records and customer behaviors were trapped in legacy, siloed formats. This fragmentation, combined with extreme traffic spikes of tens of thousands of requests per second during peak dining hours, created a massive high-concurrency engineering bottleneck. This would result in app latency, fracturing the instant, conversational user experience required for the digital storefront. 

Integrating advanced AI workloads into a heavily used live app also posed major security risks regarding Personally Identifiable Information (PII) and application stability.

To bridge these gaps, PCSC chose Google Cloud as its core strategic partner not only for its robust infrastructure capable of sub-second streaming but also for its comprehensive, enterprise-grade AI ecosystem. 

Two mobile phone screens display a Chinese AI food app showing pop-ups to select gender and dietary preferences
The "AI Food Passport" leverages consumer demographics and diverse dietary preferences (such as weight management, eco-friendly plant-based diets, and senior nutrition tags) alongside convenience store sales data centralized in BigQuery. This allows the system to deeply analyze shopping habits and latent wellness needs to deliver tailored, real-time personalized nutrition solutions.
To serve 20 million members seamlessly, we needed to eliminate data silos and scale instantly during peak hours.
The Google Cloud unified data platform and enterprise AI ecosystem gave us the secure, high-concurrency foundation required to safely turn massive operational data into real-time, personalized customer experiences.

Uni-President Chain Store Corporation

Architecting a secure AI platform with real-time streaming and grounded insights

PCSC engineered a decoupled, API-first microservices architecture on Google Cloud to transform its massive retail footprint into an intelligent ecosystem.

The foundation relies on BigQuery as a centralized data engine to ingest, standardize, and de-identify fragmented transactional logs and customer behaviors. 

To process this data under extreme high-concurrency conditions, PCSC deployed Dataflow. This fully managed streaming pipeline ensures that millions of members receive sub-second, real-time responses to conversational queries during peak dining rushes.

The team also utilized the Gemini platform to power the core conversational experience of the "AI Food Passport." To eliminate generative AI hallucinations and guarantee sound nutritional advice, PCSC implemented a retrieval-augmented generation (RAG) architecture. This setup uses grounding techniques to anchor all AI outputs directly to the verified database of the Millennium Love Health Foundation—a Taiwanese public health nonprofit ensuring all meal suggestions align with professional, safe nutritional standards.

Simultaneously, developers accelerated time-to-market by deploying an AI portal on Google Kubernetes Engine (GKE). GKE allows the team to containerize recommendation models and scale compute resources elastically, shrinking prototype testing and development cycles from months down to a few weeks.

Security was prioritized at every layer, as PCSC integrated strict access controls through Identity and Access Management (IAM). 

Combined with end-to-end encryption, the entire AI pipeline now operates in a highly secure environment that safeguards customer privacy by ensuring zero exposure to PII.

The platform yielded immediate business growth during its pilot phase, turning personalized engagement into measurable store outcomes. By delivering tailored meal suggestions at the exact moment of hunger, the "AI Food Passport" doubled the click-through rate (CTR) compared to traditional advertising campaigns. This surge in digital engagement translated directly to physical retail success, boosting the co-purchase rates of fresh food items during critical meal hours.

Two smartphone screens displaying an AI food recommendation app showing meal options with calorie counts and tags
Based on individual consumer preferences and dietary needs, the "AI Food Passport" leverages grounding techniques through the Gemini platform to anchor all generated recommendations to an expert dietitian knowledge base. By utilizing nutritional tags such as "rich in vegetables," "diverse protein," and "low calorie," the system provides consumers with customized, healthy meal plans firmly rooted in nutritional science.
By grounding Gemini models with our expert foundation database and running microservices on Google Kubernetes Engine, we successfully delivered a highly scalable, secure AI architecture that generates trusted recommendations while protecting customer privacy at every step.

Uni-President Chain Store Corporation

Redefining the OMO omnichannel retail experience through actionable data

Beyond the consumer-facing app, Google Cloud’s ecosystem catalyzed a broader internal transformation. 

By using the Gemini platform’s vector search capabilities alongside a secure RAG architecture, PCSC built a custom corporate AI assistant. This internal tool handles cross-departmental documentation queries, drastically reducing manual search times and enhancing employee productivity.

Looking ahead, PCSC is leveraging these successful recommendation models to accelerate its transition toward "OMO (Online-Merge-Offline) Situational Proactive Marketing." 

Utilizing BigQuery data sharing technologies, the retailer plans to map comprehensive cross-channel customer personas. This will allow the AI engine to proactively recommend multi-brand product bundles across the entire OPENPOINT ecosystem, expanding hyper-personalization from just dietary advice into all areas of daily digital consumption.

As PCSC scales these conversational experiences across the 20-million-strong loyalty ecosystem, unified analytical pipelines will open new monetization channels. 

By transforming operational data into high-value retail media opportunities, PCSC can help brand partners launch hyper-targeted, context-aware digital campaigns with unmatched conversion potential. 

Backed by elastic infrastructure to rapidly deploy future models, PCSC is proving that the next generation of retail isn't just about answering customer questions—it is about running an entire enterprise at the speed of real-time intelligence.

PCSC’s deployment demonstrates how generative AI can transform complex operational data into immediate business value.
By unifying data silos on Google Cloud, they have established a new global benchmark for secure, real-time personalization in the high-concurrency modern retail landscape.

Uni-President Chain Store Corporation

Uni-President Chain Store Corporation is one of the largest retail operators in Taiwan, best known for operating the region's expansive network of over 7,200 7-ELEVEN stores. Specializing in retail management, logistics, and supply chains, the company is committed to driving retail technology forward by leveraging advanced AI solutions to build innovative features across its enterprise ecosystem.

Industry: Retail

Location: Taiwan

Products: BigQuery, Dataflow, Gemini, Google Kubernetes Engine (GKE), Identity and Access Management, Gemini platform

Google Cloud