Framer AI AgentsDesign and publish professional sites with AI
Winnowly
ai, hr, screening, recruiting
9 followers
ai, hr, screening, recruiting
9 followers
Winnowly — desktop app for HR that scores and ranks resumes against a job description with AI — running fully on your own machine. Download free for Windows, unlock with a license.
The local-only setup is a nice touch for anyone worried about uploading sensitive candidate data to another cloud service. Ranking felt solid on a quick test against a job description.
@damlagt42 Thanks, really appreciate that! Privacy was the whole reason for the local-only design — candidate data should never have to touch someone else's cloud. Glad the ranking felt solid, and happy to hear any feedback if you put it through a bigger test.
The fact that it runs locally and still ranks resumes pretty accurately against a JD is impressive, no data leaving the machine. The desktop app feels snappy on my old Windows laptop.
@hlyasoyserwd24 Thanks, glad it's running smoothly! The app is lightweight by design. One thing to note — the local AI ranking works best with ~8GB of GPU memory, so on an older laptop you may want to use a smaller local model or the Claude option for the scoring step. Either way the app stays fast and your data stays put. Appreciate the feedback!
How does Winnowly actually handle parsing for scanned PDFs or resumes with weird formatting, and does it need an internet connection at all after install?
@kerimflcj Scanned or image-only PDFs currently can't be processed, as OCR isn't supported yet. These files are automatically detected, flagged, and skipped. OCR support is planned for future releases.
The application runs entirely offline using a local model. An internet connection is only required if you decide to use Claude for candidate scoring.
Does the AI ranking actually work on parsing scanned PDFs or only text-based resumes? Curious how it handles messy real-world files.
@kenan784649 It can rank any text-based resume, including those with complex layouts, unusual fonts, or written in non-English languages. However, scanned or image-only PDFs are not currently supported, as OCR has not yet been implemented, though it is planned for future releases. These files are automatically flagged as "unreadable" and skipped, while the remaining resumes continue to be processed and ranked normally.
Scored a batch of 30 resumes against a job description in under a minute and the rankings actually made sense, not just keyword matching. Nice that it runs locally, so candidate data never leaves the machine.
@tcellat38122 Thanks, really appreciate that! Moving beyond keyword matching was exactly the point — it evaluates each resume against the real requirements, not just buzzwords. And keeping it local means candidate data never leaves your machine. Glad it held up on a full batch!