Learn concepts, not tools

The belief

Frameworks endure. Tools rotate. STP (Kotler via Wendell Smith, 1956) is still live in AI-PM practice today: the concept has not changed; the tools running segmentation have changed dozens of times. Investing in the tool layer is renting; investing in the concept layer is owning.

The analogy: three types of people. A driver uses the tool. A mechanic repairs it. An engineer knows it inside-out and can create similar tools from first principles. AI makes drivers redundant. "Being technical is not about knowing a technology but using the technology" - concept-application over tool-knowledge.

How to apply

1. When picking what to study, default to the concept behind the tool. Do not learn Figma; learn design thinking and visual hierarchy. Do not learn a specific AI orchestration framework; learn how agents reason, how context windows shape output, how retrieval changes the answer. The tool changes. The concept compounds.
2. Use named frameworks as the answer architecture. When you reach for a response to a PM question, reach for JTBD, STAR, AARRR, V/V/U, NFR - not a general description. Frameworks are the evidence that a concept is internalized, not just recalled.
3. Treat tool fluency as concept-extraction substrate. "Play with APIs when you are bored" - the point is not API literacy for its own sake. The point is: each API you touch teaches you what the underlying system can and cannot do. Fluency is the input; the concept is the output.
4. Audit your learning diet for concept-to-tool ratio. A course that is 80% tool-specific (shortcuts, UI flows, vendor features) has a short shelf life. One that is 80% concept-specific has durable value. The ratio is the test.
5. When AI commoditizes a tool layer, ask what concept the tool was running. ChatGPT commoditized surface-level research. What it did not commoditize: information architecture, synthesis judgment, knowing which framework applies to which problem. Those concepts are the durable layer. Double down there.

What this is not

• Not an argument against using tools. The belief is learn the concept first, not ignore tools. A practitioner who cannot use the current toolset is not applying concepts - they are just collecting them. The mechanic metaphor permits tool fluency; it just doesn't stop there.
• Not a prescription for academic abstraction. Concepts learned without application harden into trivia. Learn STP from a professor, then apply it to a real segmentation problem, then teach it. That is the cycle. A mental model that lives only in a textbook is not load-bearing.
• Not a claim that certifications are worthless. The sibling belief certifications-are-collectibles handles the credentialing question. This belief is about the learning target - concept vs. tool - not the format.

Argues against

• "Learn the AI tool stack thoroughly - tooling knowledge is what employers hire for."
• "In a fast-moving field, tool proficiency gets you in the door faster than abstract frameworks."
• "Concepts are embedded in the tools; learn the tool and the concept comes with it."

Where to go from here

If you want to understand why breadth requires a concept-based depth-axis to stay competitive, go to breadth needs depth. That belief sets the demand; this one describes the mechanism.

If you want the practice-cadence layer - how to re-engage the same concepts until they become instinct - go to muscle memory over novelty. Concepts-not-tools is triage: what to study. Muscle-memory-over-novelty is cadence: how often.

If you want the parent theme with the full evidence arc, go to the career reflection theme.