Is Your Business Idea Any Good? An AI Sanity Check in 10 Minutes.
Your co-founder says yes. Your friends say yes. Your mother says yes. None of them is the market. An AI sanity check won't build your business — but it will tell you whether you should spend the next six months trying.
What a real business idea sanity check tests
A proper sanity check is not "do I like it" or "is the category hot." It's four concrete questions, answered against live data:
- Does demand exist? — Not in your head. In search trends, marketplace listings, forum complaints, subreddit questions.
- Is there enough of it? — TAM you can actually defend with a public source, not a made-up figure.
- Can you reach it? — A distribution channel with a realistic CAC, not hope.
- Will the math work? — Revenue minus cost times volume that fits reality.
Most founders skip three of these four and convince themselves on the first one. That's why 90% of startups fail.
What happens in a Cortex AIF sanity check
We take your 2-3-sentence idea and run the classify block. It extracts industry, geography, target customer, and revenue model — from your own text. Then the research block pulls 10-20 live sources matching those parameters. Then the operations and financial_lite blocks run the four questions above.
Ten minutes later you have a weighted score, a verdict (GO / CONDITIONAL_GO / NO_GO), and a list of the specific dimensions your idea is weak on. Not a generic pep talk.
Common sanity-check red flags we see weekly
- "Uber for X" where X has a 20-year incumbent — the TAM fight is not winnable at a startup's cost.
- Marketplaces with no chicken-or-egg answer — you need one side first, and you don't have a plan for it.
- B2B SaaS priced at $5-10/month — CAC will always exceed LTV by 3-5x at that price point.
- "AI for [common task]" without a data moat — OpenAI can ship your feature next Tuesday.
Sanity-check your idea now
Paste your idea in 2-3 sentences. Free first scan, no account needed.
Start free analysis