About Citation Check
A source-integrity platform for the age of AI, built by people who got fooled once and would rather not repeat it.
The quote that started this
While researching this very brand, we found a glowing endorsement attributed to Professor Emma Thompson at the University of Oxford. It was quotable, confident and exactly what we wanted to hear.
She does not exist. The quote was invented, and it took real work to establish that.
That is the water everyone publishing now swims in. AI writes a fluent, authoritative paragraph in seconds, complete with citations that were never real.
What we do
We check the claims in a draft against the sources behind them. Does the source exist, does it actually say what the claim says, and is it current and real.
When a citation is hallucinated, dead, weak or wrong, we say so plainly and help you find one that holds. Every result comes with the source text attached, so you never have to take our word for it.
Today we are not a writing tool. We sit between "sounds right" and "is sourced," and we close that gap.
What we will not claim
We are in a market where every competitor calls their AI magic. Being straight about the limits is the differentiator, not a confession.
- We will never say we never hallucinate, or that we are 100% accurate.
- We will not tell you a claim is true. We tell you whether the source you cited says it, which is narrower and checkable.
- We will not frighten you about misinformation to sell a subscription. The relief of a clean, sourced draft can do its own talking.
Every verdict shows the real source, the supporting text and the date, so you can check the result yourself. Our method and models stay ours.
Those do not conflict. You never need to see our engine to check our answer, and showing the work always means showing the evidence.
The enemy is unsourced information, not AI
AI did not invent the unverified claim. Lazy sourcing, copy-pasted statistics and invented experts predate it by decades.
What AI did was industrialize it. More unsourced claims, faster, dressed in confident prose.
So we use AI to fight it, and we say so plainly. The difference is direction. A model asked to be the source of truth invents one, and ours goes and reads the real document instead.
Who builds this
A small team that cares about editorial integrity more than it cares about shipping a bigger feature list. We would rather describe what the product does today and change the words when reality changes.
We believe a shared, verifiable reality is worth defending, and that it gets defended through patient work rather than grand gestures. If we get something wrong, tell us and we will correct it.
Stop trusting citations. Start verifying them.
Paste a draft. See every claim checked against a real source.