IdeaConclaveToo many ideas? Rank them best to worst, with proof.
IdeaConclave scores every idea on the factors you choose, fills them with real data, and has Claude, GPT and Gemini judge each idea blind. Then they debate, kindly, until one ranked answer wins. Every number shows its source. Every disagreement stays on the record.
How it works
Four steps from a messy idea list to a ranked list you can defend.
Dials: your factors, your weights
A Dial is one thing you care about, turned up or down. Add, remove or reweight Dials any time and the ranking updates. The AI models never see your weights, so they judge the idea, not your preferences.
| Idea | Search demand | Competition | Fit with my assets | Market size | Council |
|---|---|---|---|---|---|
| Mobile dog grooming app | 7.4 | 5.1 | 2.0 | no data | 6.5 |
| Local roofing lead site | 8.2 | 3.0 | 9.0 | 6.8 | 7.0 |
| Kids' coding comic | no data | 7.7 | 1.0 | no data | 4.5 |
Real data first
Monthly searches and competition come from Google data providers. Asset fit checks the domains you already own. Each score links to its source.
Honest about gaps
No data means "no data", never a guess. Ideas with thin evidence are pulled toward the middle so one lucky number can't win.
Your numbers count too
Type in market size, cost to build or weeks to revenue, with where you got it. Your notes travel with the score.
The council: blind first, then a fair debate
Research on AI debate shows independent answers are a strong baseline and that long debates add little. So the council is built to stay honest.
1. Blind round
Each model scores the idea alone, with its reasons and risks. This round is kept and shown, so you can see if the debate changed anything.
2. Anonymous debate
Models see each other only as "Panelist A, B, C". A moderator frames the disagreements, one panelist plays devil's advocate, and scores change only for a stated reason.
3. Forced consensus
The final score is the panel's median, worked out by the software, not picked by a chairman. You get a confidence level and the strongest dissent in plain English.
Full transcript
Read every round, reveal which model said what, and export the whole conversation.
What makes it different
| Typical AI idea validator | IdeaConclave | |
|---|---|---|
| Scope | One idea at a time | Your whole list, ranked |
| Factors | Fixed by the tool | Your Dials, your weights |
| Evidence | Model opinion | Real data with sources, gaps shown |
| Multiple AIs | Side by side, or one summary | Blind round, anonymous debate, median consensus, dissent kept |
| Cost | Subscription | Your own API keys, cost shown before each run |
Questions
How does IdeaConclave rank ideas?
Each idea gets a 0–10 score on every Dial that has data. Your weights combine them into one score. Ideas missing a lot of data are pulled toward a neutral middle, and a coverage badge shows how much of the score rests on real data.
Which AI models sit on the council?
Claude, GPT and Gemini to start, and any other model with a compatible API. Using different companies' models is on purpose: they make different mistakes.
Who pays for the AI calls?
You bring your own API keys. IdeaConclave shows an estimated cost before every run and stops at a spending cap you set.
Will it tell me my idea is great?
Only if the panel thinks so. Models are told to be kind but honest, to treat missing data as unknown, and not to invent numbers. The strongest dissent is always shown.
Is it available now?
IdeaConclave is in early development and in use by its founder. Early access opens to a small group of founders first.
Get early access
Tell us what you'd rank first. We'll reach out when a spot opens.
Email for early access