1
Dogfood first
Run the loop on your own venture or operation before you sell the method to anyone else.
Stop if: Stop if your public story outruns a lab you actually use.
Business idea engine: cultivate, compare, test, evolve.
Lab is the invention engine. Process recipes are the atomic methods for each stage. The loop improves the next question. It does not choose your purpose or prove the idea.
Integrity: be what you say you are — eat your own dogfood first. We run this loop on our own play ventures before we ask you to trust the method. Cultivate Collect hunches without forcing a pitch. Compare Put beneficiaries, constraints, and evidence side by side. Test One voluntary, reversible move that can falsify a belief. Evolve Return what happened. Keep, change, pause, or stop.
Invention system
Lab invents. Process recipes make each loop stage concrete. Skills appear on recipes only where an execution_skill already exists — there is no separate skills index.
Wire each stage to a live recipe below, form evidence on Process Artifacts, and dogfood on play ventures that name ops requirements for the shared substrate.
Set up your AI-native business lab.
Follow these steps in order when exploring value-transformation ideas. Scale effort only after market signal — not after a convincing prototype alone.
2
Name the transformation
Who benefits, from which state to which, and what must stay protected?
Stop if: Stop if the beneficiary is only "the market" or "users."
3
Justify the energy
Require repeat pain, willingness to pay or spend time, or another falsifiable demand signal before you scale effort.
Stop if: Stop if the only evidence is model enthusiasm or a polished demo.
4
Label lab vs product
Mark every work item as throwaway lab exploration or as something people may depend on.
Stop if: Stop if experiments share customer promises or production credentials.
5
Choose by reversal cost
Build first when wrong is cheap to unwind; align and spec first when wrong is expensive.
Stop if: Stop if habit or doctrine picks the method.
6
Connect direction through process
Wire Principles, Beliefs, Control, and Process before you generate at scale — purpose, assumptions, bounds, and reusable methods stay human-owned.
Stop if: Stop if agents choose purpose, consent, or stop.
7
Stand up a cheap lab
Use bounded agents, corpus, and receipts in a surface meant to be discarded — not production until promotion.
Stop if: Stop if customer data or live secrets enter the lab.
8
Run the learning loop
Cultivate → compare → test → evolve on one idea at a time.
Stop if: Stop if you skip compare or skip returning consequences.
9
Verify independently
Someone who did not build the artefact checks it; name what verification costs.
Stop if: Stop if the author accepts their own output as proof.
10
Promote or discard with a date
Cross to product only with owner, criteria, and evidence — otherwise discard and route learning deliberately.
Stop if: Stop if the roadmap gains an undated admission.
11
Return beneficiary truth
Record who benefited and how you know — separately from how fast you shipped.
Stop if: Stop if velocity stands in for value.
Inspect how we dogfood the method on live play ventures — then bring what you learn back to the next round.
Energy gate
Justify the energy before you scale.
A lab can produce convincing output quickly. That is not the same as validated demand.
Scale effort on a value-transformation idea only after a falsifiable market signal: repeat pain someone will describe in their own words, willingness to pay or spend time, or another observation that could prove you wrong.
Enthusiasm, model fluency, and a polished prototype are cheap in lab mode. Name what would falsify the idea before you treat speed as proof — then name what must stay protected in Principles.
Orientation
Which value thread is in play?
Labels for honest comparison — not a claim that every back-office function lives in this guide.
| Thread | Ask |
|---|---|
| Strategy-to-market | Who benefits and what would falsify demand? |
| Market-to-order | How does interest become a committed next step? |
| Order-to-cash | Where does money move and who keeps authority? |
| Procure-to-pay | What is bought, under what limit, with what receipt? |
| Fulfilment | What does done mean for the beneficiary? |
| Record-to-report | What state is recorded for later review? |
| Evidence-to-improvement | What returned and changed the next decision? |
| Governance | Who may stop, spend, publish, or promote? |
Which job needs doing?
Job A
Explore a new revenue stream
You have a hunch about unmet need, a possible business model, or a market shift—but you haven't validated demand or fit yet. Start here when the question is greenfield.
Cultivate a greenfield ideaJob B
Value-optimize what exists
You have an existing operation, product, or workflow. AI and complementary tech might unlock latent value, but you need to compare options and test safely. Start here when improving the known.
Optimize an existing systemThe ingredients are already here.
Dreamineering connects human-owned direction, explicit beliefs, bounded action, reusable methods, and returned evidence. Together they help people cultivate, compare, test, and evolve business ideas without pretending uncertainty has vanished.
Human-owned direction
Principles name who should benefit and what must stay protected. The engine can clarify a purpose, but it cannot choose one.
Name what mattersBeliefs exposed to evidence
Observations, assumptions, confidence, and falsifiers keep an attractive possibility from becoming an unquestioned story.
Test the mapBounded action
Control turns one uncertain choice into an authorised move with constraints, a review point, and a way to stop.
Choose the boundsReusable methods
Process contributes opportunity framing, business idea checks, experiments, evidence practices, and standards.
Open the methodsCultivate → compare → test → evolve.
Each stage links to one live Process recipe (or evidence). The loop improves the question and the next choice — it does not prove that an idea will work.
01
Cultivate
Collect frictions, slow hunches, possibilities, and early signals without forcing them into a pitch too soon.
Choose / make a strategic decision — Process strategy hub.
Open strategy recipes02
Compare
Make beneficiaries, assumptions, constraints, opportunity costs, and evidence maturity visible side by side.
Live recipe: compare prediction with consequence.
Compare outcome and correct method03
Test
Choose the smallest voluntary, reversible action that can challenge the weakest important belief.
Live recipe: one reversible experiment that can change a named decision.
Preregister a bounded experiment04
Evolve
Return consequences and surprises to the next question. Keep, change, combine, pause, or stop the idea.
Return evidence, then come back to Lab for the next round.
Open Process evidence Return to Lab
Authority boundary
The engine supports judgment; people retain it.
It does not choose your purpose, beneficiary, consent, accepted risk, spending, or stop decision. It does not prove demand, value, or improvement before evidence exists.
A useful return is a clearer comparison, a bounded test, an observation, a surprise, or a changed decision—not a confident-looking score. People affected remain able to question the frame, refuse the move, and judge whether the consequence was good.
Start with the Play page to name what you have. Use Process for the smallest fitting method, then bring what happened back to the next round.