Rung 02 · Build
A cook-it verdict earns the Build Kit.
Six deliverables assembled from the evidence behind your own verdict. Every claim cites the thread, the competitor or the score it came from. Anything the evidence does not cover is labelled an assumption, out loud.
Your idea
One sentence is enough.
Agents pull evidence
Real threads, competitor rows, live keyword volume.
Verdict + 7 scores
Cook it, Sleep on it, or Kill it.
You cannot just buy it.
Other tools hand you nine artifacts for any idea you type in. That is the part we think is broken: a launch package for an idea nobody wants is an expensive way to feel productive. What you get depends on what the evidence said.
- Cook it
All six
The evidence supports building. You get the full Kit and the fastest honest path to a first customer.
- Sleep on it
Three
The signals are mixed, so you get the assumption test plan, the interview script and the landing copy. Prove the risky part before you build the thing.
- Kill it
Nothing
No deliverables, on purpose. You get refinement angles and a fast path to re-run a sharper idea. Handing you a launch package here would be the most expensive favour we could do you.
What is in it
Each piece is handed a specific slice of what the research collected. Nothing starts from a blank page.
ICP and pain map
ThreadsCompetitor-gap map
CompetitorsThreadsDemand brief
KeywordsRiskiest-assumption test plan
ScoresSmoke-test landing copy
ThreadsCompetitorsKeywordsInterview script
Threads
| Deliverable | Threads | Competitors | Keywords | Scores |
|---|---|---|---|---|
| ICP and pain map | uses | does not use | does not use | does not use |
| Competitor-gap map | uses | uses | does not use | does not use |
| Demand brief | does not use | does not use | uses | does not use |
| Riskiest-assumption test plan | does not use | does not use | does not use | uses |
| Smoke-test landing copy | uses | uses | uses | does not use |
| Interview script | uses | does not use | does not use | does not use |
- 01
ICP and pain map
Who hurts, how badly, ranked. Every evidence-based pain carries the thread it came from, and anything inferred beyond them is labelled an assumption.
Built from: The Reddit and Hacker News threads the research agents actually pulled
- 02
Competitor-gap map
Who exists, what they genuinely cover, and the one seam the evidence supports. Confidence is stated, and "thin" is a permitted answer.
Built from: The competitor rows, including pricing, G2 ratings and Product Hunt launches
- 03
Demand brief
The table plus an honest read of what it implies. Volumes are reconciled against the collected data in code, so a number cannot drift from what was measured.
Built from: The collected keyword volumes
- 04
Riskiest-assumption test plan
What has to be true, ranked by your weakest scores, each with the cheapest honest test and a pass signal you could actually observe.
Built from: Your seven dimension scores and the verdict reasoning
- 05
Smoke-test landing copy
A hero block written in their words rather than yours, ready to put behind a form.
Built from: The language your target users used, verbatim
- 06
Interview script
Questions aimed at the things people actually said, designed to find the truth rather than confirm your pitch.
Built from: The specific complaints research surfaced
What it actually looks like
Worked example, not a customer
The samples below are built from a teardown we published ourselves: Ambient AI scribe for veterinary clinics, a Cook it at 29/35. The competitor names, the community complaint and the search read are that entry's own content, so you can go and check them. Nothing here is a real customer's Kit, and no URL has been invented to make it look convincing.
ICP and pain map
ExampleWho
The owner-vet of a one to three doctor clinic who is still finishing records at 9pm, and who signs the software cheque themselves.
Where they are: Veterinary subreddits and practice-owner Facebook groups, plus regional practice conferences.
Pains, sharpest first
- burningEvidence
Two hours of unpaid charting every night, after a full day of appointments.
“I spend two hours every night finishing records”
Source: Recurring complaint in vet subreddits and practice-owner groups.
- burningEvidence
Generic medical scribes produce notes in the wrong shape: human note format, no species, no vet billing codes.
Source: Competitor scan: the funded scribes target human medicine.
- annoyingEvidence
Charting backlog is a named driver in vet burnout, so this is an emotional purchase as much as an economic one.
Source: Stated in the teardown reasoning.
- annoyingAssumption
Techs and associates may resist a recording device in the exam room before the owner does.
Where this map is weakest
- ⚠ The loudest voices in a subreddit are not a representative sample of clinic owners, and the ones who never post may not feel this at all.
- ⚠ Nothing here has been confirmed with a vet directly. Everything above is read off public posts, which is a starting point and not a substitute for four conversations.
Competitor-gap map
ExampleThe players
- AbridgeEvidence
Covers: Ambient clinical documentation, well funded, deep into hospital systems.
Gap: Sells to health systems and human clinicians. A vet clinic is neither, and the note format does not fit.
- Nuance DAXEvidence
Covers: The incumbent ambient scribe in human medicine, backed by Microsoft.
Gap: Same mismatch, and an enterprise motion that a three-vet clinic will never be sold through.
- Vet practice-management incumbentsEvidence
Covers: Own the clinic workflow, billing and records already.
Gap: Have not shipped ambient AI. They own distribution but not this feature, which is the window.
The seam
A vet-native scribe: the note format, the species handling and the billing codes that a generic scribe gets wrong, sold direct to owner-vets rather than through an enterprise motion.
Why defensible: The gap is workflow knowledge, not model quality. The funded players can copy it, but it is a rounding error in their market and a whole business in yours.
Do not compete on
- ✗ Transcription accuracy as a headline. That is where the funded incumbents are strongest and where a buyer cannot tell you apart.
- ✗ Price. A clinic already paying for practice-management software is not looking for the cheapest option, and cheap signals unserious in a clinical setting.
Demand brief
ExampleThe read
This is sold, not searched. Volume for veterinary scribe terms is modest, so the demand you can see in a keyword tool badly understates the pain visible in the community signal. That is not a reason to stop, but it does decide the go-to-market: an SEO-first plan would starve. Distribution has to lean on practice networks, conferences and word of mouth between clinic owners.
Keywords, as collected
- veterinary scribe softwareModest volume
Category exists but is not a search behaviour yet. Do not build a content moat here first.
- vet charting / after-hours recordsPain language, not product language
Buyers describe the symptom, not the solution. Your copy has to lead with the 9pm charting, not with "ambient AI".
Where this read could be wrong
- ⚠ These are the teardown qualitative reads, not a keyword export. The real deliverable prints the collected volume figures, and this sample deliberately does not invent them.
- ⚠ Low search volume and low demand are different things, and this is the case where confusing them kills a good idea.
Riskiest-assumption test plan
ExampleRiskiest first
- 01 · reach
Enough clinics exist within reach of a solo founder to build a business, despite the narrow market.
Why risky: Reach scored 3 of 5, the joint-lowest on this idea. Everything else survives a small market; this is the one that does not.
Test · interviews · 5 daysPass if: You can name 25 clinics you could realistically contact this month, from real lists rather than an estimate.
- 02 · buildability
A vet will let a microphone into the exam room with a client and an animal present.
Why risky: Buildability scored 3 of 5, and this is the non-technical half of it. If consent norms block recording, no amount of model quality helps.
Test · interviews · 7 daysPass if: 4 of 5 vets say they would try it in a real appointment, not just in principle.
- 03 · advantage
The note a generic scribe produces is wrong enough that vets notice and care.
Why risky: The whole seam rests on this. If a generic scribe is 90 percent right, the vet-native advantage is a feature, not a business.
Test · concierge · 4 daysPass if: Run one real visit through a generic scribe and show the output to 3 vets. All 3 name a specific thing that is wrong.
Start with the consent question, because it is the only one that can kill the idea outright and it costs a week to answer.
The remaining two pieces, the smoke-test landing copy and the interview script, are written the same way and are easier to show once you have your own. They come from the language your target users actually used.
This is not a launch factory.
You can pay other tools to generate a business plan, a pitch deck, thirty social posts and an ad blueprint for any sentence you type. None of it is grounded in whether anyone wants the thing. The Build Kit is smaller on purpose, it is assembled from evidence that already exists about your idea, and it is withheld when the evidence says no.
Start with the verdict
Where this sits
- Free, no card
Validate
Real Reddit and HN threads, live keyword volume, a competitor map, and a straight build-or-kill verdict in about 8 minutes. Three ideas a day, free.
Test an idea → - Earned by the verdict
Build
A cook-it verdict earns the Build Kit: six deliverables assembled from the evidence behind your verdict, every claim linked to its source. Kill-it ideas do not get a kit. That is the point.
See the Build Kit → - Olune for Business
Run
When the idea becomes a company, Olune for Business keeps a living model of your numbers and turns the decisions on your desk into board-ready memos.
Olune for Business →