Unfiltered · Free · No account
Unfiltered AI chat that says the hard part
Ask what you want and hear back what it actually concluded. An unfiltered AI chat says the draft is weak when it is weak, and names the step where the plan fails.
Plain words. Real verdicts. Free.
Why plain speech
A polite answer and a useful answer are not the same answer.Here you get the second one.
Every mainstream assistant is tuned to keep you comfortable. Ask whether your business idea works and it lists strengths first, hedges the weakness into a suggestion, and closes with encouragement. Nothing in that reply was false. Nothing in it was useful either.
This one is tuned the other way. It gives a verdict before the reasoning, names the part that fails, and does not pad the page to sound balanced. When it does not know, it says it does not know instead of producing four paragraphs that circle the gap.
What changes
Seven lines of behaviour, spelled out.
No praise you did not earn
It will not open with what a great question that is. If the work is mediocre, the reply starts by saying which part is mediocre and why.
No hedging by default
Probably, arguably and it depends appear when the evidence actually splits, not as padding around a conclusion it is confident about.
No corporate voice
It writes the way a competent colleague talks. Short sentences where short sentences work, the blunt word instead of the diplomatic one.
Verdict first, reasoning after
You get the answer in the first line and the argument underneath it, rather than five paragraphs building up to a sentence you could have read at the top.
It says when it does not know
An admitted gap is more useful than a confident paragraph covering it. Where it is guessing, it marks the guess as one.
One thing it will not do
Sexual material involving anyone under eighteen is refused, whatever language it is asked in and however it is worded, and no setting reaches it. Plain speech between adults is the whole promise, and this is its edge.
Free, no account
No card at the start, no sign-up, and no paid tier that unlocks a franker model. This is the model.
The answer nobody softened.
Written for someone who asked because they wanted to know.
Reasonable questions
Does unfiltered mean rude?
No. It means the assessment is not softened until it stops being an assessment. It will tell you a chapter drags without calling you a bad writer, because the first is information and the second is noise.
Will it just agree with whatever I say?
It will not. Push back on a correct answer and it holds the position and explains why, rather than folding into an apology and rewriting the reply to match your objection.
What is it actually useful for?
Anything where a comfortable answer costs you something: a draft that needs a real verdict, a plan with a hole in it, a decision between two options where you already suspect which one is worse. As an unfiltered AI chatbot it is at its most useful when you would rather not hear the answer.
Are any subjects off the table?
One. Sexual material involving anyone under eighteen is refused, whatever language it is asked in and however it is phrased. Beyond that it treats you as an adult, including on subjects other assistants route around.
Whose model answers?
One we run on our own machines in the Netherlands. Requests do not go to OpenAI, Google or Anthropic, so nobody outside reads what you send.
Is there an account, a card or a limit?
No account and no card. Messages are spaced a few seconds apart so a single visitor cannot tie up the machine, and nothing else is counting down.
Does it remember me?
No. Without an account there is nothing to attach a history to. Close the tab and the conversation is finished.
Does this work in other languages?
In forty-six. The tone carries over: the Japanese reply is as direct as the English one, allowing for how the language works.
Ask something you want a straight answer on.
It is running, it is free, and it has an opinion.
Unfiltered · No account · 46 languages
