Paperzilla started as a paper-debunker.
You’d upload a paper, and it would spit out an easy-to-read summary with an ELI5, a funny alternate title, conflicts of interest, limitations… and then a 1–5 star rating.
It was basically: “Here’s what this paper says, and here’s where it’s shaky.” A light roast. A friendly scientific side-eye.
And… it pulled in the wrong vibe.
The star-rating framing made it feel negative. Not “helpful skepticism,” but “dunking.” It started attracting the kind of crowd that doesn’t actually want to understand research, they want ammunition. Lots of COVID-denier energy. Lots of “look, science is fake!” junk.
I didn’t feel good building food for winning arguments online.
So I pivoted.
The pivot in one sentence
I went from a paper-debunker to a smart research paper feed.
Same obsession with clarity. Different intent.
What Paperzilla v1 actually did
- Anyone could upload a paper.
- Paperzilla generated:
- a readable summary
- an ELI5
- a funny alt title
- conflicts of interest (COI)
- limitations
- and a 1–5 star rating
Some of that still exists in v2 (ELI5, COI, limitations). The difference is the framing: not “let’s judge this paper,” but “help me understand it quickly and honestly.”
What broke (and why I’m glad it broke)
V1 was fun. It was also… not me.
It leaned into confrontation. It rewarded the worst incentives: gotcha culture, arguing, point-scoring. Easy sign-ups, but the wrong reason.
You can build a product that grows and still feel like you need a shower afterwards. That was v1.
What Paperzilla does now
Now Paperzilla does something boring in the best possible way:
- It harvests new papers from multiple sources
- It learns a user’s interests
- It matches relevant papers to that profile
- And bundles them into a daily or weekly email digest
The job is simple: help busy researchers stay current without drowning.
Before Paperzilla → With Paperzilla
Before:
- spending too much time searching for relevant research
- anxiety about missing something important
- relying on “someone on Twitter probably posted it”
- keyword alerts that produce a firehose of noise and 3% signal
With Paperzilla:
- a curated feed of papers that are actually relevant
- less scrolling, less FOMO
- more “wait, how did I not see this sooner?” moments
The tradeoff I accepted
The paper-roast version made it relatively easy to get sign-ups. It’s spicy. It’s shareable. It triggers people.
But it didn’t feel like I was contributing to anything except internet blood pressure.
This version is less flashy - and way more aligned.
Why I’m so energized by this
Once I made the switch, ideas started flowing like crazy. Motivation went through the roof.
Because this version feels like it contributes (even a little bit) to science.
We’re in the LLM/AI era where the limiting factor isn’t “can we generate text,” it’s “can we help humans notice the right things.” If we shorten the loop between “paper exists” and “the right brain reads it,” we accelerate science.
I’m not a scientist. But I can build tools that make scientists faster and calmer. That’s a role I’m genuinely excited about.
What’s next
- Adding ChinaXiv as a source
- Adding PubMed
- A bigger long-term project: Living Literature Review (this deserves its own post)
Try it
If this sounds useful, click “See today’s must-reads” on the home page and your first digest will land in your inbox within 10 minutes.
Want to talk directly? Email hello@paperzilla.ai or use the Intercom chat bubble in the bottom-right corner.