Someone asked what Inkory is for when they already use Zotero. Fair question. I assumed the two were the same thing for most of the first year.
They are not. They work on different layers.
A paper, and a sentence
Zotero answers: who wrote this, when, where was it published, and how does it get formatted in the reference list. It stores metadata plus the PDF. Pull an item from a publisher page or a library catalogue and it keeps the author, year, volume, issue, pages, DOI.
Browser highlighting answers: what did this sentence actually say. You find a line in the third paragraph that supports your argument. Its value isn't which paper it came from, Zotero handles that. Its value is the specific wording.
Academic writing needs both at once. "Research suggests X" needs an X that came from a sentence, and that sentence needs a source that can be formatted. Metadata alone doesn't tell you what the paper argued. The sentence alone can't produce a reference entry.
Two common swaps
Assuming Zotero remembers what you highlighted. It can, on one condition: you saved that page as an item first, then highlighted inside Zotero's reader. If you read an article in your browser and closed the tab, Zotero has nothing.
Assuming a highlighter replaces Zotero. It doesn't. A highlighter captures page title, site name, publish date. That is enough for an in-text citation and not enough for a compliant entry. Journals want volume, issue, page range, DOI, fields that do not exist on a web page.
The worst version I have seen: every sentence in the body carries a parenthetical, and the reference list at the end is empty. The parentheses were typed by hand, and nobody thought about what has to come after them.
How they fit together, concretely
Zotero for the paper. Browser connector for articles, drag the PDF for everything else. Fix the metadata by hand when it is wrong (preprints, conference papers, newsletters). There is no shortcut, but you do it once.
Highlight on the page itself. Highlight while you read, in the browser. The extension stores the sentence, the page, the title, the site, the publish date. If the page changes later, the snapshot holds the text as it was.
Write both into the draft. Cite as (Author, year) and keep one matching entry in the reference list. The writing desk logs every citation you insert and, at export, lists the sources used in the piece in GB/T 7714, APA or MLA.
Reconcile in Zotero at the end. My export is plain text, so it pastes into Word. If the piece is going to a journal, I regenerate the same list in Zotero with the same CSL style and compare the two. Practical reason: web metadata is often incomplete, and incomplete metadata produces half an entry.
Three details that go wrong
Page numbers. PDF page order and printed page numbers often differ by a few pages (cover, contents, ads). Citations use the printed number, the extension reads the PDF index. No tool decides this for you.
Year. A preprint and the published version can be a year apart. A web article's last-updated date and its publication date can be a year apart. Academic citations use the publication year, not the update year.
Clickability of a web snapshot. A Zotero snapshot is a local copy, so keep the URL separately. I put the domain in the parenthetical and the URL plus access date in the reference, so the entry survives the page going offline.
What Inkory does and doesn't do
There is no Zotero integration and I am not going to pretend there is. It handles the sentence layer: highlights stay findable, they come back to you while you draft, and they export as reference entries. The metadata layer stays with Zotero.
That division seems right to me. A tool that tries to own both layers usually does neither of them well.
Related: GB/T 7714 and APA: where auto-generation should stop · Getting Readwise highlights into your draft