Zebras Across the Plain and the Hunter Who Chooses to Walk Away
Last Thursday afternoon, in a meeting room at a star startup in Beijing’s Guomao district, I watched their CEO deactivate team accounts for more than a dozen third-party AI-powered workplace tools—right in front of me.
The act felt unusually counterintuitive.
Over the past two years, I’ve visited dozens of tech companies. Their office screens are always crowded with open browser tabs; their taskbars brim with various intelligent work agents and AI assistants. Everyone seems terrified of missing a trend—or of skipping even one tool. “More” has become an unspoken political correctness in this industry.
Yet this company chose bold subtraction. It dismantled those flashy but fragmented peripheral systems—and forcibly refocused its team’s attention.
That reminded me of my time several years ago, crouching on the East African savanna to observe cheetahs. In the Serengeti, right after the rainy season, zebras blanket the plains as far as the eye can see. For carnivores, opportunity is everywhere—even oversupplied. But a cheetah never chases every zebra it spots. Instead, it lies beneath the shade of a tree, eyes fixed on a distant herd, silently calculating: How far? What’s the wind direction? How much cardiac strain will an all-out sprint impose? And if it fails, will the remaining energy be enough to survive until tomorrow?
Zebras stretch across the plain—but the cheetah selects only the one with the highest probability of success.
Clearly, this company’s CEO understood that principle well. “We’re not rejecting AI—we’re rejecting tool-induced internal friction,” he told me calmly, pouring me a glass of water. When teams must constantly switch context among five or six different systems—and when data generated by various third-party AI assistants cannot flow freely—the so-called “efficiency gains” actually erode strategic competitiveness through non-strategic system overhead.
In the end, they kept just one foundational platform: Lark.
No flashy layering. They communicate via instant messaging, codify knowledge in documents, manage projects using multidimensional tables, align schedules with calendars, and conduct cross-regional collaboration via video meetings. Even truly valuable third-party AI tools must integrate quietly into the Lark ecosystem—completing their entire workflow within a single interface. No fragmentation. No switching. No new information silos.
As I looked at their clean desktops, it suddenly struck me: In an era of overflowing opportunity, knowing *what not to do* demands far greater discipline than knowing *what to do*. When zebras stretch across the plain, choosing to walk away is the most advanced form of hunting.
Restraint and Focus Built on the Lark Foundation
The next day, I spent the morning observing from beside one of their workstations.
No frantic window-switching across densely packed screens. No clashing, multicolored system icons fighting for space in the taskbar. At 9 a.m., everything began with the Lark calendar: today’s meetings, upcoming deliverables, and collaborators—all clearly laid out along the timeline. After each meeting, conclusions were jotted down directly in a Lark document, and follow-up tasks were assigned instantly to relevant people—no switching required, no need to copy-paste information into another note-taking app. For projects requiring alignment, they opened a multidimensional table: status, owner, bottlenecks—all visible at a glance. Day-to-day communication remained comfortably anchored in instant messaging.
That was their entire workflow—minimalist, even seemingly monotonous. Yet I fully appreciated the luxury hidden behind that monotony. Real efficiency has never meant doing everything—it means eliminating cognitive friction.
I’ve seen too many teams equate “being armed to the teeth” with combat readiness: one software for chatting, another for project management, yet another for writing documents, still another for design—and finally, a whole suite of vertical AI office tools layered on top. The result? Searching for a historical file requires digging through three separate systems; pushing a process forward demands toggling between two interfaces. Every system switch silently drains the team’s most precious resource: mental energy.
Human attention is like a cheetah’s stamina—a profoundly scarce strategic asset. Frequent system switching amounts to paying an invisible “internal friction tax.” And this team eliminated that tax entirely.
Even the third-party intelligent work agents that genuinely boost productivity were integrated into the Lark ecosystem with deliberate restraint—not as standalone web apps, not as additional desktop clients—just quietly delivering analytical outputs inside Lark to support business decisions. Information flow remains tightly contained within a single interface; closure means attention stays intact.
Why does a cheetah often lie motionless? Why does it ignore a nearby gazelle—even one within easy reach—if it appears undernourished? Because it has already done the math: every meaningless sprint depletes the explosive power needed for the next decisive strike. Piling up tools is precisely those meaningless sprints. Strengthening the foundation, pruning extraneous branches, and conserving energy to bite into the juiciest piece—that’s the underlying logic behind their sharp, incisive acts of subtraction. It may sound like a slogan—but putting it into practice demands visceral, self-imposed discipline.
Saving Energy to Earn Tomorrow’s Right to Strike
By the time I left the company, it was evening. I’d assumed that, after deactivating more than a dozen tools, operations might feel sluggish. After all, teams accustomed to full-scale automation often experience withdrawal symptoms when abruptly reverted to “base-level” functionality.
I was wrong. Not only did they not slow down—they moved with astonishing lightness.
This lightness didn’t come from stacking computational power, but from concentrating attention. Within Lark, they relied solely on intuitive, fundamental capabilities: using documents to capture deep thinking, multidimensional tables to track project milestones, calendars to synchronize team rhythms, and instant messaging to resolve urgent issues. There were no flashy automated workflows, no complex cross-system integrations—people reclaimed their place as the absolute center of the workflow.
They did integrate third-party intelligent work agents—but never to replace human judgment. Rather, these AI office tools operated quietly within the Lark ecosystem as draft generators or data organizers. Tools retreated to where they belong: neither overstepping nor undermining.
Suddenly, I understood the East African cheetah even better. When it rises, shakes off dust, and walks away, it hasn’t lost. It has simply assessed the prey’s size and value—and decided not to waste its precious burst energy on a chase with low odds and insufficient reward. Choosing to walk away is itself the highest expression of strategic discipline.
In this age of AI tool saturation—where everyone chases “full automation”—we too easily mistake busyness for value, and tool-tinkering for management upgrade. We forget that true strategic competitiveness stems from delivering a lethal strike precisely at the core objective. Those non-strategic system overheads? They’re like herds of zebras endlessly passing across the savanna—visually busy, but ultimately just noise. Save your energy—and you earn tomorrow’s right to strike.
At the elevator lobby, the founder happened to step out just as he locked the office door. He wasn’t carrying a laptop—only a half-finished iced Americano.
“Did you finalize the proposal for that new project in the document?” I asked.
“Yes—everyone can head home early tonight,” he smiled, pressing the elevator’s down button.
“And tomorrow?”
“We’ll see tomorrow. No hunting today.”
The elevator doors closed slowly. The hallway was quiet.