A UC Berkeley term project · by Nico Bauer

Keeping up with the literature,
without reading all of it.

Literature Assistant takes the first pass for you. It pulls the newest papers on a topic, has an LLM read and score each one for relevance, and learns from your relevant / not-relevant calls so the next batch lands closer to what you're after.

Invite-only · Accounts are gated by a registration key.

Daily Brief — today
Ranked for your topic12 new
Deep RL for safe motion planning9/10
Graph neural networks: a survey6/10
Off-topic preprint2/10
Discover — research
Closest to your tasteround 2
Sim-to-real transfer for controltaste 0.91
Safe exploration in RLnew angle
Risk-aware motion planningtaste 0.85
Millions
Papers across journals, IEEE, ACM & arXiv
OpenAI
Scored by OpenAI's latest models
On demand
Fetch and rank whenever you want, no batch to wait for
1–10
LLM relevance score per paper

Two modes, two jobs

One for staying current day to day, one for digging in when you're actually writing.

Daily Brief

A daily check on a field: it grabs the newest papers, the LLM rates each one 1–10 with a short reason, and you skim a ranked list instead of the whole feed.

Discover

For deeper research, like working on a paper. It takes the papers you've kept and finds the closest new matches, plus genuinely new directions across the wider literature.

Built-in evaluation

Checks whether the LLM actually beats plain keyword and embedding ranking, measured against the papers you label, with real precision and ranking metrics.

How it works

From a research question to a ranked reading list in a few minutes.

Write a research question. The LLM turns it into a proper academic search query.
Fetch and rate. It pulls a broad batch of candidates and scores each one for relevance, with a short reason.
Label what's relevant. Your relevant / not-relevant marks become the answer key.
It learns. Discover refines its search from what you keep, and the evaluation shows whether any of it actually helps.

Built to make literature review more efficient

From a research question to a ranked, relevance-scored reading list in minutes. Literature Assistant handles the first pass of the literature so your time goes to the papers worth reading, not to sifting the feed.

Create an account