AI should enter the workflow, not remain a chat window
A strong answer is not the same as a reliable capability. AI becomes useful when it is placed inside a task structure with evidence, quality standards and human approval.
Working principles
- Define the intended outcome before prompting.
- Separate research, judgment, production and review.
- Automate repetitive work; keep accountability human.
- Measure reduced rework, not only faster drafts.
A zero-to-one product must pass four connected tests
Ideas do not become products through inspiration alone. Market evidence, product definition, organizational execution and real feedback must work as one chain.
Working principles
- Is the user problem specific and costly enough?
- Can the team agree on value, scope and constraints?
- Are design, engineering, supply chain and go-to-market aligned?
- Does launch feedback change the next decision?
The value of cross-industry work is judgment transfer
Breadth matters only when it helps us recognize stable variables while respecting what is genuinely different in each industry.
Working principles
- Transfer methods, not ready-made answers.
- Map the value chain and failure modes quickly.
- Respect regulation, materials and channel differences.
- Use real outcomes to correct prior assumptions.
