Literature Review Agent

Overview

Literature Review Agent turns a research question into a structured, traceable literature review by planning the process, finding and screening papers, extracting key data, and drafting a report.

Use it when you want a complete, citation-backed overview on a focused topic without manually juggling searches, spreadsheets, and notes.

Open Literature Review Agent

When to use Literature Review Agent

Use the agent when you need:

• A structured literature review report with in-text citations.
• A transparent trail of which papers were considered, included, or excluded and why.
• Screened and filtered papers matching your methodological or quality requirements.
• Extracted data ready for further analysis or export to your own tools.

Typical use cases:

• Planning a thesis or dissertation chapter.
• Writing the related work section of a paper.
• Preparing a grant application with up-to-date evidence.
• Mapping the current state of research on a niche question.

What you can provide as inputs

The agent works best when you are explicit about your question and constraints.

Core research question or topic

Start by entering a clear research question or focused topic, for example:

• "How does daily social media use relate to anxiety, depression, or self-esteem in young adults?"
• "Does personalized learning with AI tools improve student outcomes compared to traditional instruction?"

The agent uses this to design its search and screening plan.

Requirements and constraints

Optionally add requirements so the agent can screen and prioritize papers that are most relevant to your project. You can specify:

• Domain or subfield
  Examples: "clinical psychology", "urban planning", "computer science education".

• Target journals or venues
  Examples: "prioritize journals indexed in Scopus", "focus on journals like Journal of Sleep Research or Sleep Medicine", "conference papers from NeurIPS and ICML only".

• Study type or methods
  Examples: "empirical, quantitative studies only", "exclude case reports", "focus on randomized controlled trials".

• Quality indicators such as SJR or SNIP
  Examples: "prefer journals with SJR in Q1 or Q2", "prioritize SNIP above 1.0 where available".

The agent uses these requirements during the screening phase to decide which candidate papers to include or exclude and how to rank them.

Choosing sources for your review

Select where the agent should search and which papers it can consider as candidates.

You can:

• Start from the full Paperguide database (~200M papers) for broad coverage.
• Limit to PubMed when you work in biomedical or health-related fields.
• Use your reference manager when you already curated a set of papers and want the agent to screen and synthesize those specifically.

How the agent works end to end

Literature Review Agent follows a multi-step workflow to keep the review transparent and traceable.

What you get as outputs

After running the Literature Review Agent, you receive several linked artifacts.

These outputs give you both the narrative answer (the report) and the evidence trail (lists, screening outcomes, and extracted data).

Examples: questions and requirements

Use these example prompts and requirements as a starting point for your own projects.

Example research questions

Example topics and questions

• Social media and mental health — How does daily social media use relate to anxiety, depression, or self-esteem in young adults?
• AI in education — Does personalized learning with AI tools improve student outcomes compared to traditional instruction?
• Sleep and academic performance — How does sleep duration/quality predict grades, attention, and memory in students?
• Bias and fairness in AI — Which bias-mitigation methods most effectively reduce unfair outcomes without hurting performance?
• Reproducibility in science — What are the most common causes of failed replication, and which interventions improve reproducibility?
• Urban green spaces and wellbeing — Do nearby parks and green spaces measurably improve mental wellbeing and stress levels in cities?

Example requirements

You can add short requirement notes when you start the agent, such as:

• "Focus on peer-reviewed journal articles from 2015 onwards, prefer journals in SJR Q1 or Q2 within psychology."
• "Include randomized controlled trials only; prioritize studies indexed in PubMed and exclude conference posters or preprints."

How this differs from the Literature Review table/workbook

Paperguide offers both the Literature Review Agent and the Literature Review table/workbook feature, and they serve slightly different roles.

• Literature Review Agent
  Automates the end-to-end flow from question and requirements through search, screening, extraction, and synthesis into a ready-made literature review report plus traceable lists and outcomes.

• Literature Review table/workbook (/literature-review)
  Provides a customizable table-based workspace where you manually configure columns, add or import papers, and work through screening and extraction in a spreadsheet-like view.

A common pattern is:

1. Use Literature Review Agent to quickly generate an initial review and structured outputs.
2. Move into the Literature Review table/workbook to refine screening decisions, customize columns, or add more detailed extraction for specific subsets of papers.

If you want hands-on control of every column and row, start or continue in the Literature Review feature. If you want a guided, automated path to a narrative review, start with the Literature Review Agent.