Build medicine, not vitamins.
Start with a consequential human or operating problem, not a feature looking for relevance.
How I think
Clarity is a product of how a team works.
Technology does not create transformation by itself. People need a shared understanding of the problem, evidence they can trust, explicit decisions, and an operating system that can carry the work forward.
Eeda Operating System · EOSCore decision rule
When evidence is incomplete, design the next decision that creates evidence.
01
Stage 01
Observe
Learn the system before proposing a solution.Learn the people, incentives, users, workflows, and constraints. Look for what behavior reveals and what people are not yet saying.
- Output
- Shared facts and initial hypotheses
Operating principles
What I want teams to experience.
Clear enough to move. Safe enough to tell the truth. Trusted enough to disagree. Structured enough that no one has to become the hero.
Confidence is not proof.
When evidence is incomplete, design the next decision that can change what the team knows.
Leave systems, not dependency.
Distribute context, ownership, and feedback loops so progress does not depend on one person.
Hire attitude; teach skill.
Curiosity, humility, and responsibility compound faster than certainty without learning.
Framework portfolio
Methods with outputs—and limits.
Product Diagnosis Canvas
Symptoms are mistaken for root causes.
- Produces
- A system map of actors, incentives, constraints, dependencies, and evidence.
- Limitation
- Improves the problem frame; it does not replace direct user research or domain expertise.
Evidence Ladder
Decisions rely on confidence instead of learning quality.
- Produces
- A ranked view of evidence and the next responsible action.
- Limitation
- Evidence strength depends on measurement quality, context, and the cost of being wrong.
Alignment Map
Stakeholders use different definitions, goals, or decision rights.
- Produces
- Shared language plus visible alignment gaps and owners.
- Limitation
- Alignment does not mean consensus; decisions still need a clearly accountable owner.
AI Trust Framework
AI teams treat trust as an abstract value.
- Produces
- Operational criteria for usefulness, reliability, truthfulness, oversight, access, and recourse.
- Limitation
- Trust dimensions must become domain-specific tests, thresholds, and ownership.