Import Failures

Most imports just work. When one doesn’t, it usually means the source didn’t hand over the highlights we expected. Here’s how to read what happened and get unstuck.

Import Status

Every import moves through a few states:

Pending — queued, waiting to start.
Processing — we’re parsing the content and pulling out highlights. Goodreads, Readwise, YouTube, and PDF imports run in the background, so this can take a little while.
Completed — the highlights are in your library.
Failed — something went wrong before we could finish.

A completed import with zero new highlights isn’t a failure — it usually means the source had nothing new to add, or everything in it was already in your library. Highlights are deduplicated by content, so re-importing the same thing never creates copies.

Out of AI credits

Some imports come back with no highlights on them — not because there was nothing worth keeping, but because there were no AI credits left to highlight them with. Nothing is lost, and nothing needs re-uploading.

Photographed pages, saved articles, forwarded newsletters, and videos wait as they are. Becoming pulls their highlights on its own once you have credits again — whether that’s a paid plan’s allowance resetting on the 1st or a trial turning into a subscription. A large backlog is worked through over a few nights, so whatever you capture next still has credits of its own.
A video is highlighted as far as the credits went, and picks up from that same point when its turn comes round — it never re-does a stretch it already covered, and never charges for one twice. A long video can take more than one night, for the same reason a backlog does.
Want it sooner? Import the video again once you have credits: it resumes where it stopped and doesn’t charge twice for the part already done. It’s the same work, brought forward.

A credit pays for the model reading something, not for what it finds — so a page it reads and finds nothing in is still spent. See How AI Credits Work.

Common Reasons, by Source

Source	Often the cause	What to check
Goodreads	The profile URL is invalid, or your profile is private or empty (no public shelves)	Make sure your Goodreads profile and shelves are public. Becoming imports the books from your shelves and adds the highlights from its own sources.
YouTube	The video has no captions	Highlights come from the transcript. If captions are disabled, there’s nothing to pull. Try a video that shows a transcript on YouTube.
Readwise	Invalid or expired API token	Re-copy your token from Readwise and reconnect.
Kindle (extension)	Highlights aren’t synced to Amazon’s cloud	See Kindle Sync Issues.
Kindle Clippings / Kobo	Wrong or empty file	Make sure you uploaded `My Clippings.txt` or `KoboReader.sqlite`, and that the file has content.
PDF	No readable annotations	We extract real PDF highlight annotations, not scanned or flattened text. If the file has none, use manual paste as a fallback.
Photos	Pages too blurry, dark, or angled to read	Unreadable pages are skipped and the rest still land. Re-shoot the skipped ones square-on in even light — a re-shot page is a new page to read, so it costs another AI credit. See Photos.
Photos, articles, newsletters, videos	You’re out of AI credits	Nothing is lost — only the AI highlighting is missing. See Out of AI credits above.

What to Do

Retry the import. Background jobs can time out on a slow source; a second attempt often succeeds.
Check the source. Open the book, video, or file where it lives and confirm the highlights are really there.
Reconnect the account. For Goodreads or Readwise, disconnect and connect again to refresh the link and token.
Try another method. Many sources have a fallback — Kindle Clippings instead of the extension, or manual paste for a stubborn PDF.