Ten agents pull live evidence from 60+ sources, score the idea 0-100 on six factors, and put a source URL behind every claim. Nothing here is rated from training data.
Every claim audited across 410 paid reports.July 2026 citation audit · Last updated September 18, 2026 · View as Markdown
Your idea, one paragraph
A booking tool for independent physiotherapists, sold monthly, that fills last minute cancellations.
10 agents · 60+ sources
Cited score
74Launch-ready
Six factors, one number, a source URL behind every claim.
Four steps, one grid. You write a paragraph, ten agents go and read the market, the runs are cross checked against each other, and a cited number comes back.
01Input
One paragraph covering three things. No deck, no spreadsheet.
A booking tool that fills last minute cancellations
Independent physiotherapists, solo practices
A monthly subscription per practitioner
02Fan-out
Each agent owns one domain, from competitors and market sizing to regulation and hiring, and pulls live evidence from 60+ data sources.
03Converge
When two model runs disagree, the verdict reruns until they converge.
04Output
Six factors, one number, a source URL behind every claim, plus risk flags and 2-3 pivots that could lift it.
Only 17.5% reach 70+.
A source URL behind every claim
crunchbase.com
sec.gov
statista.com
The scale
One number on a 0-100 scale, cut at 40 and 70. Where it lands is the whole message.
only 17.5% reach 70+
Strong headwinds
0-39 NO-GO
The market is signaling hard against the idea as framed. Rethink the core, not the wording, before you write code.
Worth reshaping
40-69 CONDITIONAL
Where most ideas land. Positioning, target, timing, or crowding is holding it back. A sharper angle often moves the number 10-20 points.
Launch-ready
70-100 GO
Strong signals across the board. Only about 17.5% of scored ideas reach this band, where most AI validators hand out a 75-85 to nearly everything.
Six factors
The rail shows how much live evidence each factor found. It is not the factor weight, and weights are not published.
Market
Market size & growth
Bottom-up TAM/SAM/SOM and whether the market is expanding or shrinking.
Demand
Real demand signals
Whether people search for, complain about, or pay to solve this. Not a hunch.
Competition
Competitive density
How crowded the space is and whether incumbents already solve it well.
Moat
Moat durability
How easily the idea is copied, and what makes it defensible.
Economics
Business model viability
Pricing power, acquisition cost reality, and a path to margin.
Risk
Regulatory & execution risk
Legal, compliance, and operational blockers that quietly kill ideas.
Limits
Your ability to execute
It rates the idea and its market, not your ability to execute.
A guarantee
A probability read on public evidence available today.
A final verdict
Change the target, positioning, or timing and the number moves.
A low score is a direction, not a verdict. It usually means one of these, and each is fixable before you build.
01
The market is crowded
15+ competitors already doing this. You need a sharper angle than the incumbents.
02
The timing is off
The market may not be ready, or it peaked two years ago.
03
The positioning is wrong
Same product, different framing, different score. One idea tested across four angles ranged from 42 to 66.
04
The target is too broad
Narrow to a specific buyer and the demand signals sharpen.
The exact factor weights, thresholds, and agent prompts are proprietary and stay unpublished.
Every factor is scored against live, cited evidence you can open and check yourself. The verdict is auditable even though the recipe is not.
You describe your idea in a paragraph. Ten AI agents query 60+ live data sources in parallel, each covering one domain (competitors, market sizing, demand signals, community sentiment, regulation). Findings are cross-validated across models; the result is a 0-100 score with a source URL behind every claim, plus risk flags and 2-3 suggested pivots.
One paragraph in, a cited 0-100 score out, in about 60 seconds.
Every claim audited across 410 paid reports.July 2026 citation audit · Last updated September 18, 2026 · View as Markdown