Enter a topic and get candidate gaps to check — each one tied to the papers it came from. They are candidates, not findings: a handful of abstracts cannot tell you what a field has not done.
The topic is reduced to keywords and searched in Semantic Scholar’s Academic Graph, with OpenAlex as a fallback. A dozen abstracts are kept and handed to the model.
Something those abstracts leave open: a population, setting, method, timescale or outcome none of them covers, or a disagreement between them that none of them resolves. Each candidate lists the papers it rests on, and a candidate that cites no paper is discarded before you see it.
A dozen abstracts are not a field. Something can be missing from this set and settled in a paper the search never returned, in a book, in another language, or in work published last month. Every candidate is a thing to go and check.
The tools run on a fixed daily budget. When a day’s budget is used up they stop until it resets at midnight UTC, and the page says so instead of failing quietly.
No. Wonders is a research workspace for searching, organising and writing with sources; it has no gap-detection feature. This is a free tool on our website, and it is deliberately described as producing candidates to check rather than gaps it has found.
Not as it stands. Take the candidate, read the papers it names, search properly for the thing it says is missing, and write the gap in your own words once you have satisfied yourself it is real.
So you can check it. A gap statement with nothing behind it is an opinion; the papers are what makes it something you can verify or disprove in an afternoon.
Up to a dozen, from one keyword search. That is the honest scale of it, and it is why the output is called a candidate.
Type a research question. Get about ten real papers, each with one line on what it contributes. No account needed.
Give a topic and the position you want to argue. Get three drafts taking different angles — starting points to edit, never text to hand in.
Paste a DOI or type the details in. Reference-list entry and in-text citation, in the style you need.
Where any of these tools returns papers, they come from Semantic Scholar and OpenAlex, the same open databases the Wonders app searches. We keep no index of our own. The thesis-statement helper searches nothing at all — it has only the two lines you type.