PRE-LAUNCH — we build when the waitlist says go

Read any ML paper —
starting from what you already know.

Paste a paper. Get a personalized reading path that traces every prerequisite — from the papers it builds on, down to the math concepts behind them. Check off what you know; the path recalculates.

Prereq isn't live yet — once enough readers join, we start building. First 100 signups get 3 months of Pro free.

Three equations in, you're lost.

Not because you can't learn it — but because nobody tells you what to learn first. Google Scholar gives you 40 related papers. ChatGPT gives you a summary. Neither gives you a path.

How it works

STEP 1

Paste a paper

DOI, arXiv link, or title. We resolve it against the full scholarly citation graph (250M+ works).

STEP 2

See your path — as a map

Prerequisite papers and textbook concepts — KL divergence, Sobolev spaces, subgradients — drawn as a subway-style roadmap, not a flat list. The whole route at a glance.

STEP 3

Check off what you know

Click a node to mark it done. Completed steps dim; what's left stays lit. The path recalculates from your knowledge frontier — you never re-read what you've mastered.

A real path, generated by our prototype

Target: Auto-Encoding Variational Bayes (Kingma & Welling, 2013)
1— Bayes' theorem & posterior inference concept
2— KL divergence & the evidence lower bound (ELBO) concept
31977 Maximum Likelihood from Incomplete Data via the EM Algorithm paper
42006 Reducing the Dimensionality of Data with Neural Networks paper
52008 Extracting and Composing Robust Features with Denoising Autoencoders paper
62012 Variational Bayesian Inference with Stochastic Search paper
72013 Representation Learning: A Review and New Perspectives paper
82013 Auto-Encoding Variational Bayes — you're ready. target

Paper ordering: actual output from our OpenAlex-based prototype (abridged). Concept steps: from our prerequisite graph for math & statistics (expert-calibrated, refined by feedback).

Maps are not routes.

Citation-graph tools (Connected Papers, Litmaps) show you maps — beautiful, and useless for deciding what to read first. Prereq gives you a route: built on the full open citation graph, plus a prerequisite graph for math and statistics where every edge carries a verification status — expert-checked, community-confirmed, or draft — instead of being scraped and trusted blindly.

Pricing

Free

$0
  • 3 reading paths / month
  • Paper-level prerequisites
  • Visual roadmap with progress tracking

Pro

$15 / month, billed annually (early-bird)
  • Unlimited reading paths
  • Knowledge-frontier profile (check off once, applies everywhere)
  • Concept-level backtracing (math & stats)
  • "What can I read next?" — paper suggestions with the smallest gap from what you already know
  • Export (Markdown / BibTeX)

FAQ

Is Prereq live yet?

Not yet — this page is how we decide to build it. A working prototype already generates the paper paths you see above, but the product itself isn't built. Once enough readers join the waitlist, we start building; everyone on the list gets access at launch, and the first 100 get 3 months of Pro free. You'll receive launch news only — no spam.

Is this another AI summarizer?

No. We don't summarize papers — we sequence them. Prereq tells you the order in which to read things, given what you already know.

Which fields does it cover?

Papers from ML/AI, statistics, mathematics, and physics work best today. The concept layer covers mathematics and statistics first; more domains follow.

Where does the data come from?

The open OpenAlex citation graph (250M+ scholarly works), plus our own prerequisite graph for math & statistics. Edges are drafted with LLMs, calibrated against hand-verified reference topics, and labeled with their verification status — user feedback promotes or removes edges over time.

Stop reading papers in the wrong order.

First 100 signups get 3 months of Pro free.