Every AI answer is verified
before it reaches you.

This is not a chatbot. It's a governed AI gateway. Ask anything — the system checks the answer for hallucinations before you see it.

01

Ask a question in any language

Type your question in English, German, Estonian, French, Russian or Spanish. The system detects your language automatically.

02

The AI generates an answer — then we check it

Your question goes to a cloud AI model. The answer comes back, but you don't see it yet. First, the gateway generates 3 independent rephrasings and measures how much they agree.

03

Disagreement = blocked. Agreement = released.

If the rephrasings contradict each other, the answer is likely fabricated — it gets blocked. If they agree, the original answer is released to you with a verification badge.

04

High-stakes topics get extra scrutiny

Legal, medical and financial questions are held to stricter thresholds. If the system is unsure, it escalates to a second verification engine for independent fact-checking.

STD

Standard Mode

Fast verification via the internal variance gate. Every answer checked in 3–5 seconds.

~3–5s
CHA

CHA Mode

Full constraint analysis. Independent second AI reviews the answer for accuracy.

~12–18s
How to read the colors
When the system checks an answer, it marks each finding with a color. Here's what they mean — no technical knowledge needed.
[CONSTRAINT]
Blue — "This is a rule."
Something is fixed and cannot be changed. A law, a limit, a requirement. Think of it like a road sign — it just is.
"The maximum loan-to-value ratio is 80%."
[BOUNDARY]
Amber — "You've hit a limit."
Something has run out or is about to. A budget, a deadline, a capacity. The system is saying: you're at the edge.
"The daily API quota of 10,000 requests has been used up."
[INCOMPATIBILITY]
Coral — "These two things don't fit."
A conflict was found. Two pieces of information contradict each other, or something you asked for isn't possible with what's available.
"Bitcoin is not supported by SEPA bank transfers."
[SIGNAL]
Teal — "The system noticed something."
A data point or observation. Not a problem, not a rule — just something worth knowing. Like a note in the margin.
"Three out of five data sources report the same value."
[STOP]
Red — "Stop. Read this."
The most serious finding. The system is telling you: do not continue without reading this. Something important would go wrong if you proceed.
"Proceeding would violate the signed NDA."
Brightness = how sure the system is
Bright — The system is sure. Trust it.
Medium — Probably right. Worth a look.
Dim — Uncertain. Double-check this yourself.
The border glow around the answer
The whole answer card gets a colored edge. This tells you the most serious finding at a glance — before reading a word.
Blue or no glow — all clear, routine findings
Amber glow — a limit was found
Coral glow — a conflict was found
Red glow — a STOP is in this answer. Find it and read it.
The one rule to remember: If the border glows red, something in the answer says "stop." Find the red tag and read it before doing anything.