Lee Kum Kee Works with Feishu on AI Digital Employees for Craft Knowledge Transfer

Lee Kum Kee, together with Feishu, is not only passing down 138 years of traditional craftsmanship to employees but also to AI digital staff.

The collaboration between the two companies goes beyond simply integrating an AI model into the enterprise. The core lies in enabling the rich, 138-year accumulated craftsmanship and experience to be inherited not only by people but also empowered to digital employees.

Over the past two years, when many enterprises discussed AI, their first thought was often, "Should we integrate a model?" However, once truly embedding AI into operations, they realize that the model is merely the starting point. How employees find information, how processes flow, which experiences remain solely in veteran employees’ minds, and which tasks are hindered by tools, permissions, or departmental silos—these details ultimately determine how effectively AI can perform.

Lee Kum Kee's story begins precisely with these details.

Starting from an oyster sauce workshop in Zhuhai’s Nanshui township and growing into a global condiments enterprise, Lee Kum Kee has spanned 138 years and six generations, bringing its iconic bottle of oyster sauce to the world. Behind this bottle lies the true legacy: deep expertise and judgment in flavor, ingredients, production techniques, and market understanding, passed down through generations.

In the past, these hard-to-quantify "craftsmanship" skills were transmitted person-to-person. Today, with colleagues working across markets and a new cohort of "digital employees," the case describes Lee Kum Kee as exploring a new form of knowledge transfer.

From Finance and Sales to the Kitchen: AI Gets to Work

For 138 years, Lee Kum Kee's expertise has been passed from person to person. Now, this knowledge is also being taught to a group of "AI new hires."

The first changes began with seemingly minor issues.

Previously, when employees encountered financial compliance questions, they had to ask in group chats and @ relevant colleagues. Although the queries weren’t complex, delays often occurred due to time zone differences, business trips, or messages getting lost in busy chat streams.

Now, there’s an additional "new colleague" in the chat—the Financial AI Assistant, which is designed to draw on Lee Kum Kee’s accumulated financial knowledge so that questions receive answers with supporting references; buyers should verify answer accuracy against source documents in a pilot. Firms considering a similar assistant should first measure their own current response times as a baseline.

Where people once had to think, "Who should I ask?," many questions no longer require waiting for specific individuals to come online. Knowledge previously scattered across people and documents is intended to become more accessible and ready to use through the assistant.

Asking the Financial AI Assistant about risks in offline promotional activities

FoH (Foodservice on-premise) sales represent a key business scenario for Lee Kum Kee, where accumulated experience is particularly difficult to standardize and document.

Sales representatives visit restaurants daily, communicating with chefs and procurement officers. Each restaurant differs in situation and needs, making it impossible to fully rely on SOPs to guide when and how to follow up.

According to the case, Lee Kum Kee developed a FoH Sales Visit AI Assistant on Feishu to support pre-visit, during-visit and post-visit tasks; buyers should verify the exact supported actions in a pilot.

In the past, valuable insights remained trapped in face-to-face conversations. Sales reps might learn about a chef’s preferences or a store’s challenges, but such details often disappeared after the visit ended.

With the assistant, each step—from preparation and communication to follow-up—is meant to be recorded, so that judgments once dependent on individual experience can be documented and, over time, turned into reusable knowledge assets.

Using the FoH Sales Visit AI Assistant for pre-visit preparation

If the previous examples focus on internal collaboration, then Lee Kum Kee’s self-developed FunCook app brings AI directly to consumers.

The seemingly simple question—"What should I eat today?"—actually troubles many people. FunCook, as Lee Kum Kee’s cooking platform, offers recipes, inspiration sharing, and community interaction. Now, it features a new AI Recommender: users can upload photos of available ingredients or describe their taste preferences, and the AI is designed to recommend suitable dishes—capability buyers should test with their own content.

Through FunCook, Lee Kum Kee aims to bring its accumulated expertise beyond the product itself and into consumers’ daily meal decisions; whether recommendations genuinely reflect that expertise is something users will judge over time.

The AI Recommender on FunCook

"AI New Hires" Do More Than Just Answer Questions

Financial queries can be raised anytime; before sales visits, AI can help prepare materials; afterward, documentation, summarization, and follow-up steps can be assisted by AI, subject to how each workflow is configured.

These intangible "new employees" are intended to integrate into daily workflows, so that AI serves as more than a Q&A tool—acting within specific operational steps where workflows are configured to allow it.

The case attributes this potential for continued action to the surrounding collaboration platform. If an enterprise already conducts messaging, document work, meetings and process updates in Feishu, some of that information may provide context for AI agents, subject to actual product support, configuration and permissions.

Within authorized limits, agents are designed to access needed knowledge and data, understand task progress, and drive next actions—such as organizing documents, sending reminders, or advancing the task—where the platform and permissions are set up to permit this; buyers should confirm which actions are supported in practice.

The intended model is for agents to collaborate with people inside controlled workflows. A pilot should verify readable data, permitted actions, human approvals and audit records rather than assume end-to-end autonomy.

For 138 years, Lee Kum Kee’s expertise has been passed down through mentorship and senior guiding junior.

Now, the newly arrived "digital employees" also take their seat at the table of inheritance.

Before adopting AI digital staff, ask how knowledge from veteran employees will be captured and verified.