AI Summary Helper
Find methods and limitations
Structure a paper before you decide how deeply to read it.
For researchers and students
Read papers faster—
without losing the source.
Recognize papers by DOI, extract study details and limitations, and ask questions that lead back to the text.
Keep answers accountable
Source jumps make it easier to check a claim instead of copying a hallucination.
Export a useful trail
Keep notes and DOI citations ready for your literature review.
What it does
Everything a reading list should have done from the start
Save it, understand it, find it again, and actually get back to it — without another subscription holding your archive hostage.
Everything a reading list should have done from the start.
Everything a reading list should have done from the start.
You choose the model, the server, and whether anything leaves your machine at all.
The problem was never saving. It's getting back to what you saved — this is built around that part.
Every save becomes searchable, connected, and exportable — a real archive, not a growing pile.
Follow what you care about, see the good news first, and get the day summed up.
Read, listen or send it to your Kindle — whenever you have a free minute.
PDFs, like any other page
Summarize browser PDFs or files from disk like any page. “What I used” shows the words, tokens and model the AI read.
20+ languages, cross-platform
Read and summarize in 20+ languages using the extension, native side panel, or bookmarklet when the extension cannot reach a page.
AI summaries, your choice of model
Choose OpenAI, Gemini, Mistral, DeepSeek, a custom endpoint, or local Ollama for summaries that never need to leave your device.
Ghost highlights, right on the page
The AI highlights important verbatim passages so you can skim key insights and export them with the summary to Kindle, LocalSend or Markdown.
Research mode: papers recognised by their DOI
Recognise research papers by DOI and citation metadata, flag trial, review or preprint status, and extract design, sample, limitations and DOI registry citations.
Compare sources on one question
Bundle articles, label sources as peer-reviewed, preprint, news or opinion, run a lateral-reading check, and compare where summaries agree, disagree or miss context.
A knowledge graph, not just a list
See how saved articles connect through shared tags and topic similarity, even when they have no tag in common.
Your library inside Claude, LM Studio and Cursor
Claude Desktop, LM Studio, Cursor and other MCP apps can search your library, quote highlights and list what is due. You control access; nothing goes through our server.
Your library on your own storage, encrypted
Sync an encrypted library to your Nextcloud, WebDAV server, S3 bucket or local folder. Publish a searchable list with a tag graph and 3D map; we run no sync server. How to set it up · Why it matters
Notes and comments on any page
Highlight text, comment on it, or save a quoted note. Organise Notes as a list or sticky-note grid by article or tag, alongside your library.
Export to Obsidian, Markdown, Word and PDF
Export summaries to Obsidian, Markdown, Word (DOCX), OpenDocument or PDF with the title, source, tags and highlights in one note.
Questions
Before you install
Search the answers by topic, provider, privacy, archive or platform.
Do I need an account to use it?
No. Install it, bring your own AI provider key or use a local Ollama model, and start saving. An account is only needed for optional byphil Cloud access or extra Kindle/LocalSend sends.
What actually leaves my browser?
Only the summarization request goes to the AI provider you configured. RSS also needs requests to your subscribed feeds; an AI recap sends headlines and short snippets only when you request it. Search, duplicate detection, tags and recommendations run locally. See the on-device architecture →
Where is my reading archive stored?
By default, in your browser on your device. You can export it as Markdown or back it up encrypted to your own Nextcloud, WebDAV server, S3 bucket or local folder. Read the sync guide →
What happens if I clear my browser data?
Local browser data can remove a local archive if you have no backup. Export your library or configure encrypted sync to your own storage before clearing browser data or changing profiles.
Which AI providers and models are supported?
You can use OpenAI, Gemini, Mistral, DeepSeek, a compatible custom endpoint, or a local Ollama model. Bring your own key, or use optional byphil Cloud access. Learn how BYOK works →
Can I summarize PDFs, YouTube videos and RSS items?
Yes. Summarize PDFs opened in the browser or attached from disk, YouTube videos through their captions, and articles from the built-in RSS reader. Watch the workflow videos →
How reliable are the AI summaries?
AI output can be wrong and does not replace checking the source. Follow-up answers show the passage they rely on, so you can compare the summary with the original page. Learn the 10-second source check →
Which browsers and platforms are supported?
Use the browser extension, native side panel or bookmarklet. An Android companion adds links from your phone; iOS is not available yet.
What happens to my saves if I stop supporting the project?
Your archive stays exactly where it's always been — in your browser's local storage on your device. The Support Pass only gates cloud-dependent extras (byphil Cloud AI access, unlimited Kindle/LocalSend sends), never access to what you've already saved. See how Digest makes saves useful again →
Is this open source?
Yes — the extension's source is public on GitHub, under the AGPL-3.0 license. It is free for personal, educational and non-profit use; companies that want to use it outside the AGPL's terms can get a commercial license.