RefScout

Finding a real source for something you already know is true

There's a specific kind of stuck that happens while writing a paper: you write a sentence you're completely confident is true — "wireless sensor networks typically trade off latency against energy consumption," something like that — and then you sit there needing a citation for a claim that feels almost too basic to need one. You know it's true. You just can't remember which paper you originally read it in, or whether you read it anywhere specific at all versus absorbing it from five different sources over two years of a PhD.

Asking an AI chatbot to just give you a citation for that sentence is exactly the situation where hallucination shows up most — the model knows the claim is plausible (because it's true and common), so it confidently generates a citation that matches the shape of what should exist, without checking whether that specific paper does.

The more reliable way to do this is to work from the claim outward: search for the topic, look at what real papers actually say, and pick the one that genuinely supports your specific sentence — rather than starting from "I need a citation" and ending at whatever an AI hands you.

That's what Cite is for. Paste the paragraph, and it flags the sentences that read as citation-worthy claims, then pulls candidate sources for each one from live databases — not to hand you a citation blind, but so you can actually look at three or four real candidates and pick the one that says what you meant. It's a shortlist, not an oracle, which is honestly the more useful thing when you already half-know what you're looking for.

Get real candidates for the claims in your draft → Cite