AI answers fast.
OPC checks whether you're solving the right problem.
A 3-minute methodology check for one-person business ideas.
Live Demo: https://kingtmn.github.io/opc-reality-check/
Same UI for both entries — not a chat window:
My Reality → What Matters → One Question (if needed) → Next Reflection → What Would Change This View
| SEE AN EXAMPLE | Prebuilt Same Question / Reality A·B — stable, no API, 15-second understand |
| TRY MY IDEA | Your own solo-business idea (experimental demo) |
The question did not change.
The reality did.
So the path changed.
Most AI systems optimize for answering.
But some problems fail earlier:
the question is framed incorrectly,
a distinction is missing,
or an assumption closes the path too soon.
OPC works one step earlier.
It checks the direction before expanding the answer.
Before you build, check whether you're solving the right problem.
OPC does not think the whole problem for you.
It helps correct the direction, surface what matters, and leave the rest to your judgment.
Sometimes DIRECT is enough.
Simple questions should stay simple.
If this way of framing problems is useful, star the repo to follow the project.
After years working in consulting, I kept seeing the same pattern:
people often didn't lack answers.
They were answering the wrong question too early.
AI makes fast answers easier.
OPC came from that observation.
Already thinking about a solo business?
Give OPC the idea.
It won't choose your industry or write a business plan.
It will try to surface:
- what actually matters now
- one question, if the path depends on it
- what not to assume
- what to test or think about next
Sometimes the useful output is space to think:
Think before you continue.
Stopping can be the correct result.
New to OPC? Read the Usage Guide.
git clone https://github.com/kingtmn/opc-reality-check.git
cd opc-reality-check/demo
npm install
npm run devThen open the local URL Vite prints.
Personal Reality
→ what matters
→ one path-sensitive question — if needed
→ local methodological closure
→ new evidence may reopen it
Mechanism evidence — not product validation:
- offline comparative simulation —
evals/comparative/v0.7/ - isolated single-model live replay —
evals/comparative/v0.7/live-replay/ - six-Method KB regression —
evals/regression/v0.9/ - negative controls where DIRECT ≥ OPC is allowed
Simple questions should stay simple.
Not a ChatGPT bake-off. Not scientifically proven. Not production-validated.
OPC selects only lenses that materially change the current path (≤ 3).
| Lens | Core question |
|---|---|
| Running Lean | Is it real? |
| Effectuation | Where can I start? |
| Company of One | Should this get bigger? |
| Stair Step | How far should I jump next? |
| Small Bets | How much should I bet on this? |
| Reusable Leverage | What can become reusable? |
(Reusable Leverage = presentation label for internal opc_cn_reference.)
Details: methodologies/
git clone https://github.com/kingtmn/opc-reality-check.git
cd opc-reality-check/demo
npm install
npm run devcd opc-reality-check/demo
npm run acceptance
npm run buildRepo: kingtmn/opc-reality-check
OPC is deliberately incomplete.
A useful methodology should not eliminate the user's judgment.
It should improve the boundary within which that judgment happens.
Wrong direction. Right answer. Still wrong.
A correct answer to the wrong problem can still take you in the wrong direction.
OPC works before the answer.
- Demo uses lightweight deterministic heuristics (experimental)
- No real-user product validation yet
- Limited live-model replay
- Bounded six-Method space
- Not an industry oracle or success predictor
- Not a chatbot / business-plan generator
ARCHITECTURE.mddocs/MECHANISM_BASELINE_V0_7.mddocs/DEMO_PRODUCT_SPEC.mddocs/USAGE_GUIDE.mdROADMAP.mdschemas/LAUNCH_VISUAL_AUDIT.mdDEPLOYMENT_CHECKLIST.mdLICENSE_DECISION.md
Social preview: assets/social-preview-a.png · assets/social-preview-b.png
Hero motion concept: assets/hero-motion.gif
Intellectual lineage: REFERENCES.md.
opc_cn_reference is a prior-art reference case only — not a fork, translation, or endorsed collaboration.
Licensed under the Apache License 2.0.
See LICENSE.
Third-party references and prior art remain subject to their respective terms.
opc_cn_reference is a prior-art reference case only, not imported upstream expression.