Content Looks Wrong

Sometimes extracted content doesn’t look quite right. Here are the most common issues and what you can do about them.

Only the first paragraph was extracted

This happens most often with infinite-scroll sites or pages that load content progressively as you scroll. The extractor captures what’s available on initial page load, which may be just the first section.

What to try: Share the URL from a different source. If you got the link from a social media app, try opening the full article in a browser, copying the URL from the address bar, and pasting that into KeepStream instead.

Navigation menus or ads are included

The content extractor tries to isolate the article text from the rest of the page, but some sites make this difficult. You might see navigation links, sidebar content, or ad text mixed in with the article.

This doesn’t affect audio playback as much as you’d expect — the AI voice reads the main content and ad text is usually brief. But it’s not ideal. There’s no manual fix on your end; content extraction improves over time as we refine the parser.

Missing images

KeepStream extracts text only. Images, charts, graphs, and embedded media are not included. If an article relies heavily on visual content, the text extraction may feel incomplete.

AI processing on partial content

Even when extraction is imperfect, AI processing still runs on whatever text was captured. Summaries, categories, and tags may be less accurate if the extracted content is incomplete. Audio playback will reflect what was extracted.

For important articles

If an article is particularly important to you and the extraction isn’t good enough:

• Try using the direct article link (not an AMP version, Google cache link, or social media redirect).
• Try sharing from a different browser.
• If the content is available as a PDF, you can upload it directly (Study Pack feature).