Open-source research assistant
Let the right papers
find you.
PageTide learns your research profile, ranks new papers on your own computer, and keeps the ones you save in a library you own.
macOS: Apple Silicon or Intel. Windows: x64.
MIT · No account · No telemetry
A research assistant for every step of your reading
Each morning a short list of new papers, ranked against your profile, each with a one-line reason. One optional AI call previews up to 20 picks at once.
Try the profile
Change a keyword and today’s five picks re-rank. Click a keyword to cycle strong, weak and off, or add your own.
Retentive Network: A Successor to Transformer for Large Language Models
Yutao Sun et al.
A retention mechanism that offers parallel training, cheap inference and good scaling as a transformer successor.
OPT: Open Pre-trained Transformer Language Models
Susan Zhang et al.
A suite of open pretrained decoder-only language models from 125 million to 175 billion parameters.
Self-Instruct: Aligning Language Models with Self-Generated Instructions
Yizhong Wang et al.
Bootstraps instruction-following data from the model's own generations to align it with few human labels.
Emerging Properties in Self-Supervised Vision Transformers
Mathilde Caron et al.
Self-supervised vision transformers learn features that expose object boundaries and work well for nearest-neighbor classification.
Rethinking Attention with Performers
Krzysztof Choromanski et al.
Approximates softmax attention with random features, with theoretical guarantees and cost linear in length.
More papers is not the problem. Finding yours is.
About 20,000 new papers appear every day1. A handful of them matter to your work, and no one can read a feed that long. Most tools answer with a bigger feed. PageTide answers with a smaller one.
It learns a research profile from what you mark useful or not relevant, then ranks each day’s new papers on your own computer. The profile is a list of plain words you can read, correct and delete. AI is optional, and its job is to explain a paper to you, not to choose it for you. Unlike a chat-based AI research assistant, PageTide does not ask you to describe what you want each time. It reads your feedback, keeps a research profile you can edit, and turns it into paper recommendations that improve the more you use it.
Your library, your PDFs, your notes and your profile stay on your computer. There is no account, no server and no telemetry.
1 OpenAlex lists 626,339 journal articles published in September 2026, about 20,900 a day; queried 2026-10-25. Source: OpenAlex.
By the numbers
- 0model tokens to rank papers and learn your profile
- 0accounts, servers or telemetry
- 25MBdownload on Apple Silicon Macs, 17 MB on Windows
- 3paper sources: arXiv, OpenAlex and CNKI
Ranking runs on your computer, without AI. Sizes are the current release downloads.
Watch PageTide in 40 seconds
Download PageTide
Version 0.1.5. A free, open-source research assistant under the MIT license.
| Platform | File | Size | Download |
|---|---|---|---|
| macOS (Apple Silicon) | PageTide-macos-arm64.zip | 25 MB | Download |
| macOS (Intel) | PageTide-macos-x64.zip | 26 MB | Download |
| Windows (x64) | PageTide_0.1.5_x64-setup.exe | 17 MB | Download |
First launch: the app is not signed
PageTide is an independent open-source project with no paid code-signing certificate, so your system will warn you the first time. This is expected.
macOS
Unzip, drag PageTide to Applications, then right-click it and choose Open. Or go to System Settings → Privacy & Security and click Open Anyway.
If macOS says the app is damaged, run this in Terminal:
xattr -dr com.apple.quarantine /Applications/PageTide.appWindows
When SmartScreen says “Windows protected your PC”, click More info, then Run anyway. The installer is per-user and needs no administrator rights.
Requirements
- macOS on Apple Silicon (M1 or later) or Intel.
- Windows on x64.
Keys for outside services
The core (picking, the profile, the library) works offline with no key. Optional services use your own key, stored on your computer.
- DeepSeek for the AI features (previews, questions, suggestions). Optional.
- OpenAlex for a wider paper source. Optional.
- CNKI for Chinese papers, in the desktop app. Off by default; needs no key. You may be asked to slide a puzzle, and you download PDFs yourself.
- arXiv needs no key.
Contributions welcome
PageTide is a research assistant built in the open. Bug reports, feature ideas, documentation and translation fixes, and pull requests are all welcome, and small fixes count as much as new features. For a bigger change, please open an issue first so we can agree on the direction.
Open an issue Pull requests Contributing guide GitHub repository Releases