From jensen-way
Physics-of-reality decision framework. Tests whether a product, feature, or spec obeys the laws of friction, inertia, cause-and-effect, and hard constraints — producing a defensible build/pivot/kill verdict backed by evidence. Use when evaluating product ideas, feature specs, or build-vs-buy decisions.
How this skill is triggered — by the user, by Claude, or both
Slash command
/jensen-way:jensen-wayThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
A decision framework derived from Jensen Huang's framing of Physical AI — extended to product and engineering decisions.
A decision framework derived from Jensen Huang's framing of Physical AI — extended to product and engineering decisions.
"The next wave requires us to understand things like the laws of physics — friction, inertia, cause and effect. The fact that I tip that thing over, it's going to fall. When I set the bottle down, it's not going to go through the table. All of these common sense physical reasoning abilities that children have, that our pets have — most AIs don't have."
— Jensen Huang, Hill & Valley Forum 2025, interviewed by Jacob Helberg, May 3, 2025
Jensen was describing Physical AI. The same observation applies to most product plans: they operate in a frictionless vacuum — a fantasy where users adopt because you ship, teams align because you scheduled a meeting, and systems scale because the diagram says so.
The core test: if you need 10 pages to convince someone this will work, the physics are probably against you. A child should be able to follow the causal chain.
State what's being evaluated:
EVALUATING: [product/feature/spec]
DECISION TYPE: [build / continue / choose A vs B]
TIME HORIZON: [weeks / months / quarters]
Then write the full chain from "we start building" to "this matters":
1. We build [X] → because [reason]
2. Which causes [Y] → because [mechanism]
3. Which leads to [Z] → because [mechanism]
...
N. Which results in [outcome that matters]
The toddler test: read each → because. Would a child accept that explanation? If you need jargon or "if everything goes right" qualifiers, mark it as a weak link. Every weak link is where the plan assumes a frictionless vacuum — these become the focus of research.
Gather evidence for the weak links. Use available web search tools.
For broad evaluations, launch 3 research subagents in parallel:
Gravity only pulls things with mass. Is there measurable evidence people want this?
Evidence required: ≥2 sources on demand. "I think people want this" is not a force.
Users are heavy objects at rest. Moving them requires force greater than friction.
If this works, large bodies nearby will be pulled toward it.
Evidence required: name specific competitors. Search three ways before concluding "no one does this."
The bottle doesn't go through the table. What are the immovable constraints?
The software mirrors the actual communication structure, not the org chart.
Every system loses energy to heat. The question isn't "does it work" — it's "does output exceed total input, including waste heat, at 12 months?"
FORCE AVAILABLE:
- Engineering time: [X person-months]
- Budget: [$X]
- User patience: [X minutes to first value]
- Org willpower: [political capital]
- Runway: [time before results required]
FRICTION INVENTORY (each consumes force):
- [Friction 1]: [force consumed] — [source]
- [Friction 2]: [force consumed]
- TOTAL: [sum]
REMAINING FORCE: [available - total]
If remaining force is near zero or negative, the plan doesn't have enough energy to reach its destination. This is the most common failure mode — not that the physics are impossible, but that the budget runs out before arrival.
Rate each law:
| Score | Meaning |
|---|---|
| Aligned | Follows physics — evidence confirms reality supports it |
| Fighting | Soft violation — possible but consuming force |
| Broken | Hard violation — the bottle is going through the table |
Verdict rules:
# Jensen Evaluation: [Name]
*Date: [date] | Sources: [N] | Confidence: [High/Medium/Low]*
## Verdict: [BUILD / PIVOT / KILL]
[2-3 sentences grounded in physics — forces in favor, forces opposed]
## The Causal Chain
[Full chain with weak links marked]
## Scorecard
| Law | Rating | One-line evidence |
| :--- | :--- | :--- |
| 1. Demand Gravity | Aligned/Fighting/Broken | — |
| 2. Adoption Friction | — | — |
| 3. Competitive Gravity | — | — |
| 4. Hard Constraints | — | — |
| 5. Org Inertia | — | — |
| 6. Entropy & Economics | — | — |
## Friction Budget
[Force available vs. total friction]
## Key Findings
### Aligned (green lights)
- [Finding] ([source](url))
### Fighting (yellow — manageable)
- [Risk] — Mitigation: [how] ([source](url))
### Broken (red — hard constraints)
- [Violation] — Why it's immovable ([source](url))
## If We Build: Critical Conditions
1. [Must be true for the chain to hold]
2. [Must be true]
## If We Pivot: What Changes the Physics
[Specific change that removes a Broken law]
## The Single Biggest Risk
"The single biggest reason this fails is: ___"
## Sources
1. [Title](url) — [one-line summary]
[unverified].npx claudepluginhub agentoptics/jensen-way --plugin jensen-wayOrchestrates multi-dimensional business idea validation via parallel sub-agents: Lean Canvas, JTBD, market/competitive/feasibility research, SWOT/PESTLE, and a weighted scorecard with verdict.
Identifies risky assumptions for new product ideas across 8 categories: Value, Usability, Viability, Feasibility, Ethics, Go-to-Market, Strategy, Team. Rates confidence and suggests tests for startup risk evaluation.
Validates startup ideas end-to-end: KB/project/web search, manifest alignment check, S.E.E.D. niche analysis, devil's advocate inversion, STREAM 6-layer evaluation, stack selection, PRD generation.