8 Years of Digital Experience: Starbucks China Tech Team Adopts Agentic AI to Transform Retail

Interview Overview

Interviewee: Luo Jinping, CTO of Starbucks China

Interviewer: Hu Taiwen, Founder and CEO of Geekbang Technology

> “Inside Starbucks, we are planning to equip every employee with a ‘digital assistant’ within three years—possibly more than one. With such an assistant, we’ll have more time to think and innovate—something AI can never truly give.”

---

Starbucks China's Journey to Generative AI

Starbucks China, a retail industry pioneer, spent eight years completing its digital transformation and is now embracing generative AI with both enthusiasm and caution.

For Luo Jinping, the early groundwork in digitization and informatization is critical — without it, data and AI lack a solid foundation.

• Retail’s history with AI: Inventory adjustment, restocking plans, precision marketing.
• Opportunity today: Agentic AI enables transformation from point solutions to full process re-engineering.

Luo observes that Agentic AI will impact:

• Consumer-facing experiences: marketing, purchasing, and service personalization.
• Enterprise-facing productivity: collaboration and communication systems.
• Back-end operations: finance, supply chain, and internal logistics.

---

Phase 1: Digitalization 1.0 — Addition

Timeline & Initiatives

• Around 2017, Starbucks China’s digital growth accelerated.
• Partnerships with Alibaba and Tencent introduced key services like Starbucks Delivers.
• Focus: Adding new systems and features quickly to meet business needs.

---

Phase 2: Digitalization 2.0 — Subtraction

Luo Jinping’s Arrival

• Cost of “addition” rose; system complexity became challenging.
• Shift toward:
• Building in-house for unique business systems.
• Outsourcing standardized systems like ERP and HR.

Key Optimization Moves

• Unified standards for databases, middleware, operating systems:
• RabbitMQ, Kafka, Redis, and MySQL became the standard stack.
• Container management platform for elastic scaling:
• Scale resources up during peak sales events and down afterward.
• End-to-end monitoring chain:
• Issue detection in real-time.
• Uptime improved from 99.49% → 99.98%.

> Technology selection is based on Total Cost of Ownership (TCO), not just immediate expense — analyzing cost implications over 3–5 years.

---

Microservices & Data Management

• Internal systems now run on a unified microservices architecture.
• Different data types:
• Structured data flows through shared systems.
• Unstructured data (images, video) routed to big data platforms for AI processing.

---

Smarter Collaboration with Business Units

Example 1: Marketing Campaign System

• Past: Build a system for each campaign → discarded after 3 days.
• Now: A reusable drag-and-drop system enables quick campaign setup without repeated development.

Example 2: Store Development Data System

• Co-created with the Store Development (SD) team.
• Uses Location-Based Services (LBS) and big data for scientific store location selection.

> “Digital transformation is iterative — AI only helps once basic digitization maturity is achieved.”

---

Phase 3: Digitalization 3.0 — Agentic AI Adoption

• Traditional AI uses: sales forecasting, inventory automation.
• Generative AI stance: cautious in production, proactive in pilots.
• Aim: Shift focus from apps to personalized brand experience via AI agents.

---

Future Customer Journey Vision

• AI recommends the best Starbucks store each morning.
• Parking spot reserved → seat allocated.
• Coffee served, tailored to customer taste, without manual action.

Transition to this vision focuses on a 3–5 year transformation roadmap.

---

AI Proof of Concept Stage

• Many projects are PoC due to privacy governance.
• Most mature: OPS Chatbot—handles internal Q&A.

---

Starbucks AI Investment Categories

• Business Value Investments
• ROI projections over 3–5 years reviewed by Technology Investment Committee.
• Productivity Investments
• Cost savings quantified with finance analysis.
• Infrastructure Investments
• Foundation work — may not show immediate value but is critical.
• Risk & Compliance Investments
• Mandatory risk-related investments with minimal internal resistance.

---

AI Productivity in Engineering

• AI programming tools: Initially optional, now significantly impactful.
• Internal tests show up to 50% of code generated by AI in new builds.
• Engineers and product managers must:
• Clearly articulate requirements.
• Communicate effectively to optimize AI output.

> “Clear expression of needs will become a core technical skill.”

---

Transitioning Internet Talent into Retail

• Many Starbucks China tech staff come from internet backgrounds.
• Key differences:
• Retail product cycles are longer & irreversible.
• Higher trial-and-error cost.
• Decisions require deeper risk-ROI analysis.

User-Centered Build Process

• Clarify “Who is the user?” and how processes will change before developing systems.

---

Organizational Culture & Structure

Human Touch

• Coffee tasting ritual before meetings enhances team culture.

Career Development Reform

• Introduced T Series technical track.
• More job grades, faster promotion cycle.
• Growth mindset prioritized in hiring.

---

Luo Jinping’s Leadership Philosophy

• Proactive engagement with business units to discover needs.
• Think from business perspective before proposing solutions.

---

Conclusion

Luo’s optimism for technology’s role in retail is grounded in:

• Permanent consumer needs (coffee culture remains timeless).
• Real-time tracking of trends.
• Partnerships with tech firms to enrich the “third space” coffee experience.

---

For aligned AI innovation, platforms like AiToEarn官网 demonstrate how content generation, multi-platform publishing, analytics, and monetization can parallel retail’s push for scalable, integrated systems. Starbucks China’s path to Agentic AI mirrors these cross-industry efficiencies.