How it works.
You bring an idea. The engine pressure-tests it against real market signal, scores it on fixed dimensions, and returns a verdict you can act on — escalating only as far as the idea earns. Here's exactly what happens at each step.
Five steps. Each one earns the next.
The engine starts free and shallow, then goes deeper — but only on ideas that survive the step before. You never pay to dig into an idea it already told you to drop.
Every tier carries everything before it.
So the only question worth answering is what the next one adds. That's all each column shows — not a grid of ticks where the tier with the most separately-nameable parts looks like the best buy.
- →Verdict & scoreProceed · Caution · Pivot · Kill, out of ten
- →Five-dimension breakdown
- →Competitive realityWho already does this
- →Napkin economicsWhether the arithmetic survives
- →Critical assumptions
- →Evidence baseEvery claim traced to a source
- →A 48-hour test
- →Beachhead ICP & personasWho to sell to first
- →Positioning wedge & brand
- →Market sizingTAM / SAM / SOM, rescoped
- →Unit economicsMargin, CAC, LTV, payback
- →Scenario modelBase · bull · bear · worst
- →Risk register
- →Voice of the marketVerbatim quotes, clustered
- →Pivot optionsIf the answer is no
- →A 30-day validation plan
- →Architecture recommendationWhat to build it with, and why
- →Product definition & MVP scopeWhat ships first, what waits
- →Feature backlogRICE-scored, ordered
- →User flows & user storiesThe whole product, written down
- →Screen blueprints
- →Legal & compliance
- →A 30-day build sprintDay by day, with gates
- →Offer design & pricingWhat you sell, at what price
- →MessagingPillars and headlines, per segment
- →Channel strategyWhere the first customers come from
- →Metrics planOne north star, with kill triggers
- →Financial refinement · Gate 2
- →Content kitLanding pages and first posts, written
- →Investor assetPitch deck for venture-scale markets, bootstrap memo otherwise — based on your own market sizing
- →A 30-day launch calendar
Not one model — a panel of agents.
A chat reply is one confident guess. Each whatifnowai run sends your idea through a structured pipeline where specialized agents research, score, and argue with each other before anything is signed off.
Research
Agents pull live signal from real sources — communities, search trends, video, the open web — specific to your idea.
Score
Each dimension is rated 1–10 against fixed criteria and weighted into a single number. No vibes — a rubric.
Cross-examine
A skeptic agent argues the bear case and flags unproven assumptions; conflicting findings get reconciled, not averaged.
Verdict
The synthesis becomes a four-tone verdict, a score, and a designed report — with every claim traceable to its source.
Sizes the opportunity and reads demand & timing.
Maps direct & indirect rivals; finds the open wedge.
Pulls real voice-of-customer to validate the pain.
Pressure-tests pricing, CAC, margins, break-even.
Designs channels, offer, and the launch plan.
Argues the bear case; flags every weak assumption.
Every agent cites its sources. The verdict is the synthesis of the debate — not a single model's opinion.
Every claim is sourced. Every source shows its rank.
Most tools run your idea through one prompt and call recycled guesses "validation." We pull live signal and label exactly how much each finding can be trusted.
One number. One word. Both earned.
Each dimension is scored 1–10 and weighted into a single headline score. That score maps to one of four verdicts — a clear call, not a maybe.
Color is data, never decoration. The same warm scale runs through every report — you can read a verdict at a glance, then check exactly how it added up.
A chat reply guesses. We show our work.
| One AI prompt | Quick validators | whatifnowai | |
|---|---|---|---|
| Real market data | None — recycled training | "50+ sources" (often broken) | ✓Live Reddit, Hacker News, YouTube |
| Sourced claims | No citations | Unreliable | ✓Every claim cited + tiered |
| Verdict you can act on | A paragraph | A score | ✓Four-tone call + 48-hr plan |
| Goes past "good idea?" | No | Stops at the market | ✓Through to a launch plan |
| Cost | $20–200/mo | Subscription | ✓€39–279, pay per depth |
Still wondering something?
Most of it is answered in the FAQ — pricing, data, privacy, how it differs from a chatbot. If yours isn't there, ask it here. A founder reads every one.
That's with us.
We'll reply to the address you gave. If it turns out to be a question other founders ask too, it ends up in the FAQ.
Stop guessing.
Start building.
Paste your idea, get a free signal check — then run the full SCAN when you're ready.
Get a free idea check →