Brutally honest AI: ask for the truth, get it

By Candor editorial team · Updated · First published

Short answer

A brutally honest AI tells you what it actually thinks instead of padding every answer with praise and disclaimers. Candor is built that way: ask it to rate your CV, your pitch or your chapter and it names the weak parts first, with the fix. It answers health, money and law questions with numbers instead of a referral. It's for adults 18+, and new accounts get 200 free credits with no card.

You paste your business plan into a chatbot and get back “This is a great start! Here are a few things you might consider…” Three paragraphs later you still don’t know if the plan is any good.

Why AI flatters you

Researchers call it sycophancy: models trained on human ratings learn that agreeing with people and praising their work scores well, so they do it even when it’s wrong (Sharma et al., 2023). The result is the compliment sandwich, every time.

What brutally honest looks like

Ask Candor “Rate my CV. Don’t be nice.” and you get:

Same with a pitch (“Who would actually pay for this?”), a chapter (“Where did you get bored?”) or an offer (“My boss offered 3%. What do I say to get 15%?”).

Honest about hard topics too

Health, money, law and relationships are where mainstream assistants hedge most. Candor answers with numbers and specifics: what a symptom usually means and when it’s worth a doctor’s visit, how much debt is too much at 30, what your lease actually lets your landlord do.

Prompts that get real honesty

Most AIs drift back to politeness. These keep the answer useful:

Honest vs harsh

Honest feedback names the problem and the fix. Harsh feedback only names the problem. Ask Candor for both halves: “tell me what’s wrong and rewrite the worst paragraph.” If the tone is more than you need, say “same verdict, kinder delivery” and it adjusts without softening the substance.

Where honesty saves you money

Start

Open candorai.dev in any browser. 200 free credits, no card. You must be 18 or older.

FAQ

Why are most AI assistants so agreeable?

They're trained to please. Feedback-style training rewards answers people rate highly, and people rate flattery highly, so models learn to praise first and hedge everything.

Will Candor be mean for the sake of it?

No. Honest isn't cruel: it tells you what's weak and why, then how to fix it. Ask it to be gentler and it will.

What can I ask it to review?

Your CV, a cover letter, a business plan, a pitch deck outline, a chapter, an email you keep rewriting, a salary offer.

Is it free?

New accounts get 200 free credits once, with no card. Subscriptions start at $19.99 a month.

Ask the question you actually have.

Free plan: chat and 1 image, no card. Chat, images and video in your browser.

Try Candor free See how it works: honest feedback

Sources

Keep reading

More in Uncensored AI