---
source_url: "https://elicit.com/solutions/search"
title: "Paper Search | Elicit: Al for scientific research"
mirrored_at: 2026-08-20T15:02:24.484Z
host: elicit.com
cited_in_42a: true
mirror_canonical: "https://index.42a.ai/elicit.com/solutions/search"
---

> **Original source:** https://elicit.com/solutions/search

## How does Elicit's AI search work?

### Ask the right question

Asking the right question is half the battle. Elicit helps specify broad research questions and points out ways to clarify your analysis.

### Rank all papers in our database by similarity to your query

We then use custom embedding models to find the papers which are most likely to be relevant to your question, including looking at features like citation count and recency.

This means we primarily rank papers by similarity in content or meaning, not keyword frequency.

### Deeper review of top 1,000 papers

We take the top 1,000 papers and use LLMs to prioritize further based on relevance to your research question.

Depending on your Elicit plan and workflow, we will then show you anywhere from the top 4 to the top 1,000 most relevant papers.

### Natural language screening

You have the option to run custom screening questions over these papers to narrow down further.

## How does Elicit's AI search work?

### Ask the right question

Asking the right question is half the battle. Elicit helps specify broad research questions and points out ways to clarify your analysis.

### Rank all papers in our database by similarity to your query

We then use custom embedding models to find the papers which are most likely to be relevant to your question, including looking at features like citation count and recency.

This means we primarily rank papers by similarity in content or meaning, not keyword frequency.

### Deeper review of top 1,000 papers

We take the top 1,000 papers and use LLMs to prioritize further based on relevance to your research question.

Depending on your Elicit plan and workflow, we will then show you anywhere from the top 4 to the top 1,000 most relevant papers.

### Natural language screening

You have the option to run custom screening questions over these papers to narrow down further.