Last month we shipped the Feed. Now we're turning it into a workflow.
The old Paperzilla feed was good for many users, but it still had a real problem: too much noise.
If a paper was off, you had two options. Accept it, or keep tweaking your interest description and hope you landed in a better place. There was no real feedback loop.
That is what changed.
As a first step toward better recommendations, the in-app paper feed now lets you give explicit feedback on each recommended paper: Not for me, Useful, or Very useful.
This came directly from user feedback. When I asked what feature people wanted next, the most common answer was simple: improve the feed by letting users send explicit thumbs up and thumbs down signals.
What changed in the app
Users can now work through the feed as a workflow.
- Mark a paper as
Not for me - Mark a paper as
Useful - Mark a paper as
Very useful
The useful part is immediate, even before the backend catches up. You can already filter out irrelevant papers more easily and build a short list of the papers that are actually worth your time.
Before, you tuned your setup and hoped for the best. Now, every vote is an explicit signal.
Why this is stage one
The workflow is live. The backend that learns from those signals is not.
I chose to ship the feedback loop first because it is already useful on its own. In my own use, it makes the feed easier to work through: remove the irrelevant papers, keep the strong ones, move on.
What's next
- Gather feedback on the workflow and improve it
- Implement the recommender backend
Read the guide: docs.paperzilla.ai/guides/feed-in-app
Email me at mark@paperzilla.ai with feedback on the new workflow. I read every message and always reply.