first-time founders, builders who code with AI
Where can I get brutal, honest feedback on my startup idea?
5 min · 2026-11
By Onur Keser, cofounder of nommi · reviewed 2026-11-08
From several places, and each one is honest about a different thing. A roast tool or a blunt mentor is honest about the logic of the idea. A public forum is honest about how strangers react to your pitch. Only the people who have the problem are honest about whether they want it. Get the hard read first, so you know which question to take to real people.
run roast my idea on your own idea →
In short
- Answer: use a roast for the flaw, a forum for first impressions, and real people for the verdict.
- Check these three things: did the feedback name one specific flaw; did it say why that flaw matters; does it point to a test you can run.
- Watch out: harsh is not the same as useful. A put-down with no reason teaches you nothing.
- Do this next: get one blunt read of the idea, then put its biggest flaw in front of people who have the problem.
- Last reviewed: 8 November 2026.
Why is honest feedback so hard to get?
Because almost everyone you ask has a reason to be kind. Friends do not want to hurt you. Colleagues do not want an awkward lunch. A general AI chat assistant leans toward agreeing with the person asking, which the piece on what to do when an AI says your idea is great covers with its source. So the idea gets praised, and the flaw that will sink it stays hidden until money and months are spent.
The options, side by side
| Where | What it is honest about | What it misses |
|---|---|---|
| Friends and family | Whether they like you | Almost everything else. They answer to be supportive, not to be accurate. |
| A general AI chatbot | Facts, market words, ideas to explore | Your idea's weak spot. It tends to agree with the person asking. |
| A roast tool such as nommi's Roast My Idea | The flaw it judges most serious, and a test for it | Whether people will pay. Only a test with real people answers that. |
| A forum or subreddit built for roasting startups | How strangers react to your pitch in a few seconds | Depth. Most replies judge the wording, and the loudest reply is not the most informed. |
| A mentor or accelerator office hours | Patterns from ideas they have seen fail | Your market's details, and their time is limited. |
| People who have the problem | Whether the problem is real and painful | Speed, and sometimes candor. Ask about what they did, not what they think, and their answers become the verdict. |
What makes feedback useful, not just harsh
Useful feedback has three parts. It names one flaw, not ten. It says why that flaw could kill the idea. And it points to something you can check this week. "Nobody will want this" fails all three. "Small gyms already track members in a spreadsheet they do not mind, so switching has to save them real time; ask five owners how long check in takes now" passes all three.
nommi's Roast My Idea is built for that shape. It returns a verdict of Build it, Sharpen it or Kill it, the biggest flaw it sees, and one test to run. If you want the full picture rather than the single worst flaw, the piece on which validation tool to use sorts the options by job.
Where people come in
Every road ends with people who have the problem. Ask what they did the last time it happened, not whether they like your idea. The piece on asking questions that get the truth shows how. The piece on why your friends' yes is not evidence explains why the people closest to you are the worst judges.
Common mistakes
1. Shopping for a kinder opinion. If the first honest read hurt, a second tool that likes the idea adds no evidence. 2. Fixing the pitch instead of the flaw. Strangers who dislike your wording are giving you copy notes, not a verdict on the business. 3. Taking every point at once. Pick the flaw everything else depends on and test that one. 4. Arguing with the feedback. Ask what would change the reader's mind, then go and find out.
FAQ · 3 questions
Is there an AI tool that roasts a startup idea honestly? Yes. A roast tool is built to find the weak point instead of agreeing with you, as a blunt mentor or a roasting forum can. It names the flaw and a test, but it cannot replace the test.
Is a roasting subreddit enough? It is good for a fast reaction to your pitch. It is weak on depth, and it cannot tell you whether people will pay.
What should I do with brutal feedback? Pick the one flaw that would kill the idea and test it with people who have the problem this week.
Stop asking yourself
If the question is still open, you are the one person who cannot answer it honestly. Run roast my idea and get the read before the building starts.