Notion AI summarizes your notes. Elicit reads papers you haven't. Different jobs.

A friend was preparing a thesis lit review and did it in Notion AI, because that's where her notes lived. She'd paste PDF text, get tidy summaries, and build an outline from them. It looked productive. Then her advisor asked for the actual findings of one cited study and the summary hadn't captured it — the model had summarized the abstract's vibe, not the paper.

That's the failure mode Elicit is built against. It pulls from a corpus of real papers, extracts the specific claims, methods and sample sizes, and shows you the snippet it came from. When I asked it what the literature says about a niche intervention, I got a table of studies with columns I'd have built by hand over a week. Not vibes — extractions, with sources attached.

Notion AI is not bad; it's just aimed at your material. For 'what did I conclude last quarter' or 'clean up these meeting notes,' it's great. It has no special relationship with the academic literature, and summarizing a pasted PDF is a coin flip between extraction and embellishment. The tell: it never shows you where in the paper a claim came from.

The workflow that made both happy: Elicit for discovery and extraction — find the studies, pull the findings, check the snippets. Then the distilled notes go into Notion, where Notion AI is genuinely useful for organizing them and finding connections across my own reading.

If you're writing anything that will be graded, reviewed or cited, the stakes decide this one. Notion AI hallucinating a finding into your draft is your problem to catch. Elicit at least shows its work. For everything that isn't research — Notion is fine. For the literature itself, use the tool that was built for it.