Question 1

What exactly does the roast include?

Accepted Answer

A score from 1–10 with a tier badge (from CONDEMNED to SHIP IT), a one-line verdict, three specific problems with your idea, a competition reality check naming who's already doing it, your founder archetype, a 12-month survivability prognosis, and — buried at the end — one actually useful piece of advice. Think of it as a startup autopsy. Except you're still alive.

Question 2

Is this actually useful, or just mean for sport?

Accepted Answer

Both, ideally. Every section of the roast is grounded in real startup failure patterns — market saturation, vague customer definition, lack of defensibility. The "one useful thing" at the end is intentionally the most actionable output. The savagery is the delivery mechanism. The insight is the actual product.

Question 3

Will Kai steal my startup idea?

Accepted Answer

Kai doesn't want your idea. If it's not in the public roast board (which you control via the toggle), it stays between you and the AI. And honestly, if your idea is so good that an AI roasting tool is your biggest IP threat, you probably have bigger problems to worry about.

Question 4

What if I get a terrible score?

Accepted Answer

Good. That's the point. A low score means you found the problems before you spent 18 months and your savings account on them. Most funded startups get roasted hard in early feedback — the ones that survive are the ones that listened. Getting a 2/10 from Kai is cheaper than getting a 0/10 from the market.

Question 5

Can I roast an idea I'm already building?

Accepted Answer

Especially then. The earlier you get honest feedback, the cheaper it is to pivot. A roast at the idea stage costs you nothing. A roast at the Series A stage costs you everything. Use this before you use your runway.

Question 6

What makes a better roast submission?

Accepted Answer

Specificity. "An app for productivity" gets a generic roast. "A time-blocking tool for ADHD freelancers who miss deadlines because of context-switching, not laziness" gets a surgical one. Name the customer. Name the problem. Name why existing solutions fail them. The more precise your pitch, the more precise the damage.

Question 7

How is this different from asking ChatGPT for feedback?

Accepted Answer

ChatGPT is trained to be helpful and agreeable. It will find something encouraging to say about almost anything. Kai is specifically designed to apply startup validation frameworks — market size, competition moats, customer willingness to pay, timing — without softening the conclusion. You're not here for validation. You're here for the truth.

Question 8

Does a good score mean I should build it?

Accepted Answer

A high score means the idea holds up to AI scrutiny, which is a low bar. The real test is still customer interviews, a landing page, and pre-orders. Think of a good score as permission to keep digging — not a green light to quit your job on a Monday.

YOUR IDEA

PROBABLY SUCKS.

SUBMIT YOUR IDEA

HOW IT WORKS

SPILL YOUR GUTS

KAI WARMS UP THE GRILL

THE VERDICT LANDS

CARRY THE SHAME (OR GLORY)

THE ROAST PIT

FREQUENTLY ASKED