From Individual Pioneers to an Enterprise Traffic System
Every time I visit enterprises for exchanges, I love posing one question to executives: “How many of your employees are secretly using AI tools?”
The audience usually exchanges knowing smiles and nods in unison.
What’s most abundant today isn’t AI tools—it’s frontline “AI pioneers” emerging organically across businesses. Yet the hardest challenge has never been *whether* tools exist, but whether they can truly be implemented.
When individuals use AI, it’s like driving a car yourself: floor the accelerator, and you zoom ahead—fast and solo. When enterprises adopt AI, however, it’s like managing an entire city’s traffic system.
No matter how many skilled drivers you have, a fully functioning traffic system won’t spontaneously emerge. If all you’ve got are lanes without traffic lights—and drivers without a central traffic management center—you’ll inevitably end up gridlocked at intersections. Many enterprises’ AI implementation stalls precisely at this “traffic jam.”
Teams integrate various third-party AI assistants and intelligent work Agents, and employees enthusiastically adopt them at isolated touchpoints. But this is merely a celebration of point solutions—not a triumph of systematic value. Data sits idle in systems; knowledge remains siloed within departments; processes break down at critical nodes.
We’ve reviewed numerous enterprises’ real-world implementation status and found that roadblocks consistently cluster around four areas: management gaps, dormant knowledge, process breakpoints, and talent misalignment. A handful of tech-savvy employees alone cannot bridge the chasm of enterprise-scale collaboration.
Letting AI handle transactional tasks—and shifting human focus toward judgment, collaboration, and value creation—is where true breakthrough begins.
Next, we’ll explore how to build this enterprise “traffic system” leveraging Lark’s everyday capabilities: instant messaging, Docs, Multi-dimensional Tables, and Calendar.
Management Enablement: Empowering Executives to Build Business Agents Themselves
Executives often confide backstage: “We’ve bought piles of AI tools—so why does business still feel unchanged?”
I usually respond with just one question: Have managers themselves rolled up their sleeves and started using them?
Enterprise AI implementation falters most when executives act only as “overseers”—standing on the shore, directing others to swim.
I once accompanied the CEO of a retail enterprise on store visits. His past inspections resembled “incognito investigations”: checking product displays, scrutinizing hygiene standards—leaving store managers anxious and staff overwhelmed. Later, he decided to try a different approach.
Together with his team, he integrated pre-configured third-party intelligent work Agents into Lark Docs and instant messaging to create a collaborative “store inspection assistant” workflow. Now, even before stepping into a store, the assistant has already compiled a clear, concise summary of that location’s recent foot traffic trends, inventory alerts, and key issues flagged in previous inspections.
Inside the store, he no longer fixates solely on shelves. Instead, he first chats with the store manager about recent developments and gauges team morale—then jointly reviews operational bottlenecks with the store manager, guided by AI-generated insights.
Inspect people first—then inspect stores. With that shift in perspective, store managers feel seen and supported—and motivation naturally follows.
When leaving, he simply creates a to-do item in Lark and directly assigns the agreed-upon corrective actions to the regional manager. No cumbersome approval workflows. No delayed manual reports. Everything flows seamlessly, right at his fingertips. This approach delivers clearer oversight than his old incognito visits—and moves faster than traditional hierarchical reporting.
When managers personally step into the arena and reshape their management lens with AI, the first green light illuminates in the enterprise traffic system.
Knowledge Enablement: Awakening Dormant System Data to Speak Up
Shifting managerial perspectives resolves the “people” bottleneck. Yet many enterprises face another silent pain point: data abundance without actionable knowledge.
During one presentation, a dairy farm owner vented onstage: “We’ve invested in multiple systems—milking, feeding, environmental control—all packed with data. Yet none of it actually guides day-to-day operations.”
Data sitting idle in systems is digital overhead; knowledge actively applied in business is digital assets.
So how do we awaken dormant data and get it to speak?
At this dairy farm, a post-2000 employee named Xiao Yu accomplished something that earned high praise—even from veteran veterinarians. He manually extracted key metrics from several systems, structured and organized them in Lark Multi-dimensional Tables, and documented analytical logic in Lark Docs.
He built a “Cow Health Profile” Multi-dimensional Table, linking each cow’s feed intake, rumination frequency, and milk yield. Then, he connected it to a third-party AI assistant, enabling the intelligent work Agent to analyze these complex metrics daily.
Whenever any cow’s metrics deviate abnormally, the AI assistant instantly pushes targeted alerts—including preliminary diagnostic suggestions—to the relevant Lark group chat.
In the past, veterinarians only spotted problems after cows fell ill. Today, Xiao Yu intervenes proactively—armed with early warnings from Lark. That’s the power of knowledge enablement.
Vast amounts of business knowledge accumulated in Lark are no longer static SOPs locked away in filing cabinets—they’re dynamic, fingertip-ready ammunition. When Docs, Multi-dimensional Tables, and third-party AI tools interoperate, data transforms into experience—and experience becomes productivity.
Let AI handle data搬运 (data transport); shift human focus to judgment, care, and value creation.
Only when knowledge flows freely can an enterprise’s moat deepen meaningfully.
Process Enablement: Embedding Time-Honored Craftsmanship into Automated Workflows
Knowledge may now flow—but if it remains stuck at the “reference-only” level and fails to embed into actual business workflows, organizations will still hit a fatal “process breakpoint.”
While researching a large steel mill, I met a veteran steelmaking master craftsman nearing retirement. His biggest headache wasn’t producing subpar steel—it was squinting through reading glasses every day, cross-checking process parameters across stacks of paper forms and legacy systems.
Decades of hard-won craftsmanship resided entirely in his head. Once he left, that expertise would vanish. So how could this craft be preserved?
Young engineers at the mill built a new workflow using Lark—firmly anchoring accumulated expertise directly into operational processes.
Now, the master craftsman simply uploads his routine inspection notes to a Lark Doc. Engineers then use Lark’s Open Platform to securely push those inspection details to integrated third-party AI office tools, retrieving AI-assisted analysis recommendations.
The master craftsman reviews the AI’s suggestions and manually enters his confirmation or key parameter adjustments directly into a Lark approval form.
There’s no arrogant assumption of full automation—only grounded, human-AI collaboration. The master’s craft isn’t replaced by machines; it’s steadily embedded into automated workflows. When intelligent work Agents connect meaningfully to real business scenarios and run stably across core processes, knowledge transfer ceases to be an empty slogan posted on a wall.
In the past, retiring senior employees took irreplaceable skills with them. Today, they leave behind standardized, live workflows running in Lark. Data flows across systems; experience accumulates within processes.
Time-honored craftsmanship shouldn’t fade with retirement—well-designed processes let seasoned expertise grow wings for the new era.
Talent Enablement: Measuring Creators in the AI Era with New Metrics
Even the sharpest tools must ultimately be wielded by people.
While accompanying enterprises on their AI journeys, I uncovered a harsh truth: many companies’ AI transformations fail because they “measure new talent with old yardsticks.”
You equip employees with cutting-edge intelligent work Agents—yet evaluate them solely on how many forms they fill out daily or how many workflows they execute. It’s like giving a race car driver a top-tier supercar—but insisting they drive the entire race in first gear.
I witnessed truly insightful practice at a cosmetics enterprise.
They launched a hands-on “executives-in-the-classroom, employees-in-competitions” initiative. Executives didn’t stay holed up in offices reviewing reports—they personally learned logic frameworks and built workflows. Employees, meanwhile, shared efficiency hacks in Lark Docs and instant messaging—like leveling up in a game—competing to solve business pain points most elegantly.
When managers lead by example, employee creativity ignites authentically.
We’ve analyzed numerous roles and found that routine, low-value tasks dominate daily workloads. In the past, employees drowned in such tasks. Now, it’s time to redefine role requirements.
Let AI handle transactional work—and shift human focus toward judgment, collaboration, and value creation. People—not tools—are the enduring engine powering AI implementation.
Throughout this journey, the Lark Services team acts more like enterprise coaches. We don’t deliver cold, rigid systems—we partner with enterprises to rebuild talent standards and measure the authentic value each creator contributes.
From management, knowledge, and processes to talent, these four pillars of enablement complete a closed loop—from tools to people.
To help you implement this framework effectively, I’ve prepared a free resource: *The Enterprise AI Implementation Playbook*. Reply with the keyword “Playbook” in the backend to download it instantly.
Point solutions empower Agents to *do* tasks.
Lark empowers people and Agents to *work together*.
Lark Services empowers enterprises to *achieve sustained results*.