Plain answers about what 4CITE does, what it does not do, and how to use it.
The 4CITE engine reads a document and detects errors in logic and manipulative rhetoric at both the surface level and the structural foundation underneath as well. The engine output includes a numerical score, a graphical display of “integrity-at-a-glance” that we call a kite, and an evidence array of supporting data.
4C-123 is the serial number for the system standard that governs the 4CITE engine and other surfaces that bear its mark.
No. 4CITE measures structure, not truth. A document can score high and still get the facts wrong. A low score does not mean a document is false or that its author lied. It means the reasoning in the text does not support its claims as well.
No to both. It does not check facts, look up citations, or guess who wrote a document. It reads how the argument is built, regardless of whether a person or machine wrote it.
Each gate asks one question about the document:
The model answers each gate in a separate call, so no gate sees another gate’s answer. More on the four gates.
We combine them with a geometric mean. That means one low gate lowers the total more than a plain average would. The total falls into one of four tiers:
A kite is the chart of one reading. Each gate gets one point: logic on the left, rhetoric on the right, surface on top, foundation on the bottom. The farther a point sits from the center, the higher that gate scored. We never turn or flip the frame, so you can compare any two kites.
The Ruler is on our home page. It shows thirteen public documents scored on 4C-123, from the Universal Declaration of Human Rights to a 1954 tobacco-industry advertisement. We scored each one alone, and the Ruler shows where each one landed. Your own score uses the same 0 to 100 scale.
The standard is pinned: its prompts, rules, scorer and sampling have been fixed since September 9, 2026, on rented model weights. When we move it onto weights we host ourselves, the Ruler has to reproduce there, and any gate that does not is retuned. Then the standard is frozen permanently.
They are the three steps, in order.
Any text document: a brief, a filing, a policy, a letter, an essay, or an answer from an AI. It must be at least 600 characters and no more than about 60,000 characters.
We store it encrypted in your private vault. To score it, we send the text to the company that runs the model. We do not use your documents to train any model, and we have not authorized any provider to do so. We do not sell your personal information. Certificates are confidential unless you share them; if someone looks up a confidential certificate’s ID, they learn only that it exists. The Privacy Policy names every provider we use.
A kite costs 5 credits. A certified kite costs 30, six kites, so 225 credits covers 7 certified kites. Full details are on the Pricing page.
For one document, get a $4 certificate. You need no account. For regular use, pick a plan on the Pricing page. 4CITE.ai is in beta.
Use the contact page.