How it works
Enter a claim, then add one or more pieces of evidence. Each piece of evidence is scored on semantic overlap with the claim, presence of a source URL, whether it contains numbers, and the credibility of the source domain. The chain score combines the weakest link with the average.
Why this matters
AI assistants build their answers from evidence chains, not isolated sentences. A claim that can be traced through 2-3 strong sources is dramatically more likely to be cited than a standalone claim, even a true one. See our how AI assistants choose citations guide.
Ship it
- Every non-trivial claim needs at least 2 supporting sources.
- Prefer primary research URLs over secondary aggregators.
- Include numbers in every piece of evidence you can.
- Pair with the Citation Builder and Citation Reliability Score.