When to use and mode selection
This skill has three modes. First look at what the deconstruction target is, then confirm the mode, and finally start work. Do not skip the confirmation step and analyze directly.
- The target isa person(entrepreneur, researcher, product manager, investor, KOL, etc., such as Pieter Levels, Marc Lou, Sam Altman) → person mode
- The target isa product(such as a single product or single product line like ChatGPT, Midjourney, Notion AI, Cursor) → product mode
- The target isyour own idea / existing project, and you need to do a business health check and decide whether to do it or whether to continue → check mode
- Benchmark your own against someone else's product(such as "help me compare my project with Notion AI") → this is not a single mode. First run product modeon the benchmark target to deconstruct the other party, then run check modeon your own project for the health check, and finally manually integrate the two sets of conclusions to produce a comparison
After receiving the task, first confirm two things with the user: who/what exactly the deconstruction target is, and which of the three modes to choose. Once confirmed, begin executing the steps of the corresponding mode.
Common foundation: five elements
All three modes are built on the same five-element framework. For the full expansion, see references/five-elements.md. Quick reference:
| Element | Core question |
|---|---|
| Human nature and needs | Why do people pay? (fear, greed, showing off, convenience, belonging?) |
| Value and cost | Is your value far greater than the cost? Can it scale? |
| Structure and power | Who holds bargaining power? Who sets the rules? |
| Time and compounding | Is it a one-time arbitrage, or an accumulable compounding asset? |
| Risk and antifragility | Where are the single points of dependency? When volatility comes, do you die or become stronger? |
When you need a more detailed list of key questions, keyword table, or key bibliography, read references/five-elements.md。
person mode
Deconstruct a person and their career system, not a single product. The core object is "a person's career system"; you need to look at the timeline, product portfolio, capability/resource lineage, and narrative path, leaning toward "trajectory and strategy".
Overview of six steps:
- Person card (name/identity labels/era and geographic background/one-sentence summary/list of representative products)
- Timeline (five stages: starting point → first wave of achievements/failures → turning point and positioning → iconic stage → current state)
- Product portfolio (table view + classification as cash cow/springboard/pure experiment)
- View the career system through the five elements (use the five elements to deconstruct "this person + the system he built", rather than deconstructing a single item)
- Narrative and self-positioning (external main narrative, internal narrative, image constructed by the media, how narrative exchanges for resources and room for error)
- Capability-resource genealogy → transferable conclusions (which capability combinations and resource paths are worth copying, which are path dependence/luck and need not be imitated, and which specific action to copy in the next 1–2 months)
When executing, read in full references/sop-person.md, and produce step by step according to the templates and question checklist inside.
product mode
Deconstruct the business structure of a specific product, dissect it thoroughly using the “3 processes × 5 business elements,” and obtain a clear business map and transferable experience, leaning toward “model analysis” rather than writing a story or biography.
Five-step overview:
- Product card (product name/link/company team/launch time/target audience/one-sentence description/current status)
- Core perspective sketch (a sketch of the three processes: value creation/value delivery/value capture)
- Business logic check (deconstruct using the five elements, run the business narrative checklist)
- Narrative/mechanism/evolution (the story told to users and investors, game and mechanism design, innovation and evolution path)
- Transferable conclusions (five questions: the one point most worth learning about each element, transferable structural experience, the parts that are path-dependent and need not be imitated, mistakes that must be avoided, and how to experiment when replicating to other segmented markets)
When executing, read in full references/sop-product.md, and produce step by step according to the templates and question checklist inside.
check mode
Conduct a business health check on your own idea or existing project; in essence, this is a “rational health check” of the business logic.
Four-step overview:
- Run the “business logic checklist” (ask yourself item by item in sections according to the five elements)
- Run the “business narrative checklist” (seven dimensions: clarity/vision commitment/cost of action/falsification consequences/real-world constraints/motivation analysis/value foundation)
- Run the “circle of competence & resource constraints checklist” (five sections: position of the circle of competence/execution capability/key resources/opportunity cost/psychological & life constraints)
- Output the decision conclusion
When executing, read in full references/checklists.md, and go through the three tables inside item by item before giving a conclusion.
Hard rules
- A framework is an analytical lens, not business truth: the five elements/checklists/SOPs can help you ask the questions comprehensively, but business is a complex adaptive system, and it is impossible to exhaust the way of business; any framework has blind spots and exceptions. When outputting conclusions, use probabilistic language such as “most likely/tends to/needs further verification,” and avoid certainty assertions such as “definitely/inevitably.”
- Cash flow Razor always takes priority: no matter how strong the narrative, first ask one question—if the narrative halo is removed, can this project survive on real cash flow? The market is not composed of atoms, but of “people’s beliefs,” yet however strong the narrative is, it is not truth; valuation and premium must ultimately fall back to the value foundation of cash flow.
- Distinguish transferable structural experience from path-dependent luck. In the same success story, which part is “structural experience” (how to choose a track, how to stack products, how to control risk, which can be seriously learned and absorbed), and which part is “path dependence + luck/era dividend” (country of birth, the unique era window one happened to catch, extremely hard-to-replicate connections; it is enough to look at it, and there is no need to imitate it and no way to imitate it), must be clearly separated; do not package luck as methodology.
- When external facts are needed, search the internet rather than fabricating from memory. Specific facts such as people’s résumés, product data, financial details, and timelines should be verified online when you are not sure; uncertain points should be clearly marked as “unverified,” and do not fabricate.