Talking with AI: How Do You Survive the AI Chaos?

The bigger the storm, the pricier the fish—but only if you survive long enough to bring it ashore. Written for everyone feeling anxious in the AI wave.

Prologue

Over the past few months, our official-account editor has been using AI to help with their work, and that practice has gradually converged on the form of Claude Code + skills + CLI. A few days ago, using Claude Code + ATA MCP/skills + OpenCLI, the editor built a personal daily hot-topic content push tool.

Screenshot of the daily hot-topic push tool built with Claude Code

And the editor got today’s article pushed by the very tool they had vibe-coded. The author says this piece came out of a conversation with AI, and with permission we’re sharing it here.

Screenshot of the push tool recommending this article

If this article helps you, I hope you won’t just bookmark it—but, starting today, go dig a moat of your own.

1. First, See the Situation Clearly: What Is AI Actually Replacing?

AI won’t lay you off directly, but it will make your boss feel the team doesn’t need this many people. A team where one P8 used to lead 10 people—now AI makes that P8 feel 5 is enough. You weren’t replaced by AI; you were replaced by the combination of “AI + a colleague.”

Big-tech layoffs aren’t random. The decision chain: strategic contraction → cut business lines → surplus headcount → rank by “irreplaceability” → trim the bottom. What the AI era changes is this: the bar for personal irreplaceability has risen sharply.

A more brutal truth: AI’s core value proposition is “doing more with fewer people.” It’s not “pivot and you’ll survive”—it’s that the total is shrinking. The pace at which new roles are created may not keep up with the pace at which old roles disappear.

2. Survival Logic from Three Vantage Points

2.1 The Executive View: Talent Demand Shifts from a “Pyramid” to a “Diamond”

The P5 layer disappears outright. Executives increasingly prefer to keep “people who can scout the path for them,” not just “people who can execute.”

2.2 P8-P9: The Most Anxious Layer

The pressure P8-P9 faces: from above, demands to cut costs and boost efficiency; from below, low AI-tool adoption; sideways, the neighboring team already has an eye-catching demo. The key: P8-P9’s own seat isn’t safe either.

2.3 P7: The Most Dangerous “Middle-Layer Squeeze”

P7 salaries run 50%-100% above P6, but they hold little power and control few resources. In many domains, AI’s depth has already surpassed P7. P7 survival strategies: become a “tech-business translator,” become an “operator of AI deployment,” and master “uncodifiable knowledge.”

2.4 P6: The Most Dangerous—and the Most Opportunity-Rich

  • 25-year-old P6: the core strategy is speed. Find an “AI+X” direction as fast as possible; you can afford to lose, so you can bet big.
  • 30-35-year-old P6/P7: the core strategy is leverage. Combine the domain knowledge you already have with AI; your tacit knowledge is your biggest asset.
  • 35-plus P7/P8: the core strategy is irreplaceable connection. Your value lies in being the node for certain key connections within the organization.

3. The Four-Layer Model of a Personal Moat

Layer 1: The Efficiency Moat (Survival Line)—Within 6 Months

How fast and how deeply you use AI decides whether you survive this round. Become the team’s “AI-native” benchmark, build a personal AI workflow SOP, and quantify your efficiency gains. But beware the “efficiency paradox”: once everyone is using AI to boost efficiency, efficiency itself is no longer a moat.

Layer 2: The Cognition Moat (Competition Line)—6-12 Months

Efficiency can be imitated, but deep understanding of the business and the technology can’t be copied quickly. Become a domain expert, develop the ability to “define problems,” and build commercial intuition.

Layer 3: The Relationship Moat (Growth Line)—Continuous Effort

Your value = your capability × your visibility. Manage upward so your boss knows you’re thinking, exert influence sideways to become a cross-team connector, and build a personal brand. Layoff decisions aren’t purely rational.

Layer 4: The Asymmetry Moat (Ultimate Line)—Long-Term Effort

Find the intersection of “only you can do it, AI can’t do it, and the organization badly needs it.”

4. Identity Crisis and Psychological Shock

When the coding ability you once took pride in becomes AI’s basic skill, you’ll go through several stages: denial → anger → fear → acceptance → reconstruction.

The cognitive adjustment: decouple “technical ability” from your sense of identity. A carpenter’s value lies not in being able to swing a hammer, but in understanding structure. Tools can change; the essence of the craft does not. Set aside at least 2 hours a week not to “produce” but to “think.”

5. Plan B Isn’t Giving Up—It’s Being Rational

A person with only one plan actually has no plan. The following signals say it’s time to accelerate Plan B: the business line adds no new headcount for two consecutive quarters; your direct boss has frequent 1:1s with HR; the team takes on an “AI replaces the existing process” project; performance reviews hint that you “need to find a new value point.”

6. Pitch-Deck-ify the PRD: From “Writing Docs” to “Telling a Story”

Anyone who can write a pitch-ready PRD has a founder’s mindset: the ability to read the industry, define problems, tell a story, and plan strategy. Put together, that’s a “product-minded technologist.”

7. Nine Underlying Insights

  1. Anxiety itself isn’t the enemy—stagnation is.
  2. AI box-ticking is wasting your most precious time.
  3. Over-optimizing the individual can hurt the team (the prisoner’s dilemma).
  4. The highest-order moat is “making the team stronger because of you.”
  5. Salary compression may arrive even earlier than layoffs.
  6. “Embrace AI” gets put in the weekly report but produces no actual effect.
  7. Most people won’t act—doing one thing beats thinking about ten.
  8. Sometimes “leaving” isn’t failure but switching tracks.
  9. In any era, the ultimate moat is “the ability to learn, adapt, and act fast in the face of change.”

A Final Word

In the AI era, your value no longer depends on what you can “do,” but on what you can “decide to do” and on persuading others to do it with you.

The deeper moat is this: when your current moat is breached by AI too, you can find a new one within a week. Don’t be a person with a fixed moat. Be a person who can always dig a new one.

OpenCLI: https://github.com/jackwener/OpenCLI