A brutal AI gives you blunt, unsoftened feedback instead of compliments. HAITE is built to do that: it leads with the biggest problem in your work, rates it honestly and tells you how to fix it.

Yes. The free plan has a token allowance instead of a message count. Paid plans give longer replies, bigger models, documents, pictures and video.

At the top of the brutality dial, yes. Lower settings are blunt without the swearing. You choose the setting.

No. The dial changes how much comfort it holds back, never whether the criticism is true. It goes after the work, never anyone's race, religion or other protected trait.

HAITE

The brutal AI that tells you the truth

Most AI assistants are built to be agreeable. Ask them to rate your plan and they find something to praise first. HAITE is a brutal AI. It reads what you give it, finds what's actually wrong, and says so - then tells you how to fix it.

Try HAITE free

A real example

Someone sent HAITE this, with the brutality dial on 10:

"handing in my notice tomorrow to go full time on tiktok. 2,300 followers, posting 3x a week. one of my vids hit 40k views last month so the algorithm clearly likes me. gonna sell a £49 'how to grow on tiktok' course. got £600 saved, rent is £750. rate it"

HAITE's reply opened: "That's not a plan, that's a f***ing hobby with a landing page." It pointed out that one 40k video means the algorithm liked that video, not them; that a course nobody has asked for isn't a business; and that they'd be £150 a month short before food. Then it said what to do instead: keep the job, keep posting, and only quit once TikTok covers the rent for three months.

Score: 2/10 - "A hobby with a landing page and a negative bank balance."

That's the point of a brutal AI. A friend would have said "go for it". HAITE did the maths.

The brutality dial: 1 to 10

You set how hard it goes. At the low end it's quiet and flat - no pep talk, just the problems. In the middle it's blunt. At 10 it swears, it's openly contemptuous, and it still answers the question properly. The dial changes how much comfort HAITE holds back. It never changes whether the criticism is true.

What you can throw at it

Essays and assignments - paste it, add the marking criteria if you have them, and get every weak argument and lazy sentence pointed out. It can also check whether your writing reads as AI-written.
Code - what's broken, what's fragile, and what a reviewer would flag.
Business plans and ideas - the numbers, the assumptions, and the part you've been avoiding.
TikToks and YouTube videos - send the link. HAITE reads the caption, looks at the cover picture and reads the spoken words when the site has them.
Anything you want rated - ask it to rate something out of 10 and it gives you a score card with the number and one honest line.

Is it just mean?

No. Mean is easy; useful is the hard part. HAITE leads with the most important problem, explains why it matters, and says what to do about it. It goes after the work and the decision, never after anyone's race, religion, sexuality, disability or background. And it isn't always right - it's an AI, and it can be confidently wrong, so check anything that matters.

Questions

Is HAITE free?

Yes. The free plan runs on a token allowance rather than a message count. Paid plans add longer replies, bigger models, Word documents and slide decks, and pictures and video.

Does it swear?

At the top of the dial. Turn it down and it's blunt without it.

Is it an honest AI or a brutal AI?

Both - it's honest at every setting, and the dial decides how brutal. If you want the honesty without the theatre, see honest AI.

Open HAITE