Validate before
you build.
One prompt. 30 seconds. A research-backed verdict on whether your idea is worth building.
Not just a number.
A complete diagnosis.
Roast Score
A single 0–100 score across 8 weighted criteria: market demand, timing, problem urgency, scalability, moat, revenue clarity, customer access, and feasibility.
Strengths & Weaknesses
See exactly where your idea excels and where it falls short. Each criterion comes with specific evidence and reasoning — not vague platitudes.
Market Intelligence
Real competitor names, funding rounds, market size, and timing signals — sourced from live web search, not stale training data.
Community Feedback
Real ideators tell you if they’d pay, flag the biggest risk, and identify what’s missing — human signal no AI can manufacture.
MVP Build Prompt
Score 40+? Generate a platform-optimized prompt for Replit or Lovable with DB schemas, feature specs, and design direction. Included in Pro plan.
Common questions
Two layers. First, AI with live web search analyzes your idea across 8 research-backed criteria — from market demand to technical feasibility, using real competitors, funding data, and market trends. Then, real people review your idea: would they pay for it, what's the biggest risk, and what's missing. AI gives you data. Humans give you signal.
No. Each of the 8 criteria is weighted based on real startup failure research (CB Insights, Y Combinator data, Bill Gross's Idealab study). The weights reflect why startups actually fail — not theoretical frameworks. A score of 65 means something specific and defensible.
After your AI roast, other ideators on IdeaRoast review your idea and give structured feedback — would they pay for it, what's the biggest risk, and what's missing. You can review others' ideas too. The more you review, the more visibility your idea gets. It's human validation that AI alone can't provide.
You just saved yourself months of building the wrong thing. A low score isn't a dead end — it's a diagnostic. You'll see exactly which dimensions are weak (market? timing? moat?) and get specific recommendations on what to fix or where to pivot.
If your idea scores 40 or above, we generate platform-optimized build prompts for Replit and Lovable. These aren't generic descriptions — they include database schemas, feature specs, user flows, and design direction. Paste it into your builder of choice and start shipping.
Because building an MVP for a fundamentally broken idea is a waste of your time and money. If the market doesn't exist or the timing is wrong, no amount of good engineering will save it. Fix the foundation first.
Yes, by default all roasts are private. You can optionally make your Roast Score public to share with co-founders, investors, or on social media — but the full analysis stays private unless you choose to share it.
Four things: (1) Live web search grounds every score in real market data — not stale training data. (2) The scoring framework is calibrated against real startup outcomes, not generic optimism. (3) Real people review your idea too — ChatGPT can't tell you whether someone would actually pay for it. (4) The output is structured and actionable — a score, a diagnosis, human feedback, and a build-ready spec. Not a wall of text.
About 30-60 seconds. The AI runs multiple web searches to find competitors and market data, scores your idea across all 8 criteria, and generates your full analysis — including a witty roast, strengths, risks, and a specific recommendation.
Stop guessing. Start knowing.
Every week you spend building the wrong thing is a week you didn't spend building the right one. Find out in 60 seconds.
Free · No signup · 60 seconds