Conclusion Generator

It drafts the paragraph that closes a piece of writing — the part that gathers the argument back up and decides how to leave the reader. The useful ones work from the finished text rather than from a description of it, because a conclusion is defined by what came before it.

Yes. Working one piece at a time runs on the free plan. Longer documents, several drafts in a single pass, and keeping the run so every piece you finish gets the same treatment are covered by the paid plans.

Most conclusions do four things in some order: restate the claim in words that are not the introduction's, pull the supporting strands into one line of reasoning, say why it matters beyond the page, and then make a closing move. The fourth is the one people skip, which is why so many endings feel like they stopped rather than finished.

It depends on the piece far more than on any rule: a blog post can close in two sentences, a research chapter may need a section with its own subheadings. A reasonable default is a tenth or so of the piece, adjusted for how many threads you actually have to bring together.

Mechanically, yes — pulling the threads of a finished piece into a closing paragraph is the kind of work a model handles well. What it cannot do is decide whether that ending is the right one to leave your reader with. That judgement is about your piece and it stays with you. If you are submitting academic work, your institution's rules on AI assistance apply regardless of how a draft was produced.

The others take an excerpt and return a paragraph, and every session starts blank. Here the whole document goes in, and the moves you accepted last time shape what comes back this time — so the tenth conclusion arrives closer to something you would have written yourself.