How Product Managers Can Thrive in the AI Era: 3 Survival Rules and a Practical Workflow

1. Mindset Shift: The Fermi Level Redefines Product Manager Value

With AI tools like ChatGPT, Midjourney, and Deepseek‑R1 automating PRD writing, prototype generation, and business logic inference, traditional execution tasks are rapidly standardized. However, insight discovery, cross‑department coordination, and long‑term strategic decisions remain challenging.

The "Fermi level" analogy, coined by Tian Yuandong, suggests that a person's value now hinges on whether they can enhance AI capabilities rather than merely matching AI output. Product managers must therefore generate combined output that exceeds AI alone.

When AI can produce a complete PRD draft in minutes, the half‑day effort of polishing a document loses value. The core value moves from delivering standardized artifacts to three AI‑hard‑to‑replace dimensions:

Ability to pose critical questions – uncover hidden user pain points or anomalous data signals that define problems AI can truly solve.

Deep decomposition of complex problems – break vague business goals into clear, logical instruction sets that AI can execute, turning wishes into actionable commands.

Design and trade‑off under constraints – craft optimal solutions considering resources, time, business goals, and user experience, a process beyond AI’s probabilistic combinations.

Product managers stuck at merely collecting, translating, and passing on requirements risk falling below the Fermi level, producing high‑efficiency but low‑value AI‑generated content.

2. Tactical Implementation: The "Review‑Deep Dive" Workflow

This workflow combats AI‑induced complacency and is suited for drafting complex PRDs, designing key product flows, or conducting critical decision analyses.

Stage 1 – AI Draft Generation : Define a precise problem and issue a clear prompt to AI (e.g., ChatGPT, Claude, Deepseek) to create a first‑draft PRD covering overview, user flow, core logic, and exception handling.

Stage 2 – Deep Review Mode : Open a new document and critique the AI output. Instead of editing sentences, hunt for logical jumps, weak arguments, and hidden assumptions. Example: if AI suggests auto‑refund for credit‑worthy users, ask what metric defines "credit‑worthy" and how to mitigate fraud risks.

Stage 3 – Deep Dive Thinking : Select one or two core contradictions uncovered, then spend at least 30 minutes without digital distractions, using paper and pen to explore the problem’s essence, stakeholders, data validation methods, alternative solutions, and long‑term impact.

Stage 4 – Reconstruct the Output : Transform the AI draft into a "Product Decision Memo" that includes annotated reasoning, justification for key decisions, and risks the AI missed.

The essence of this workflow is to force AI’s output to become an input that triggers deep personal thinking, preventing the manager from becoming a mere "AI echo chamber".

3. Strategic Elevation: From Employee to Founder

Tian proposes that every product manager will eventually transition from an employee role—focused on KPI execution—to a founder role—focused on sustainable value creation.

Build a "Founder Dashboard" centered not on DAU or GMV but on three meta‑questions:

What concrete success does my user achieve because of my feature? (e.g., reduced monthly report preparation time leading to a promotion)

How does my work build an uncopyable business moat? (e.g., unique data acquisition, deep algorithmic optimization, or ecosystem‑level UX)

How am I feeding and evolving my "AI副脑" (AI assistant) with proprietary data, structured domain knowledge, and reusable decision models?

Using this dashboard, frame AI prompts around ambitious, specific goals—such as increasing next‑month retention by 5 %—and ask AI to simulate intervention strategies, then design experiments with validation metrics.

4. Advice for New Product Managers

Beginners often over‑rely on the latest AI tools for efficiency, neglecting the essential early‑stage deep‑thinking practice. Manually sketching chaotic flows, conducting rough user interviews, and wrestling with detail‑level decisions build the intuition needed for long‑term product success.

Treat AI as a "super intern" who knows a lot but can hallucinate; your job is to assign tasks, review output, and correct mistakes, thereby sharpening your own ability to ask the right questions, evaluate answers, and chart work paths.

In a flood of AI‑generated solutions, the true competitive edge lies in formulating unique, grand‑scale aspirations and possessing the resilience, wisdom, and execution power to realize them.