SciSpace Academy
TutorialLiterature review › Step-by-step AI literature review

Mastering Literature Review: Discover, Analyse, Present with AI Agents

SciSpace AcademyEnglishLiterature reviewStep-by-step AI literature review

TL;DR

1-min read

This tutorial shows how SciSpace Agent can support a literature review from the first question to a finished, shareable report. It is aimed at researchers and graduate students who want faster, more rigorous reviews and structured outputs, while still verifying sources and shaping the conclusions themselves.

Key takeaways

  • Deep Review Mode helps you frame precise, researchable questions; a good review question is specific about context, concept and outcomes, and frameworks such as PICO or SPIDER can help structure it.
  • Multiple databases can be searched at once, with results collected into a single Combined Paper Table so that all candidate studies sit in one place for screening.
  • Findings can be filtered, organised and analysed interactively, making it easier to synthesise insights and spot research gaps and emerging trends across the literature.
  • Structured, citable reports with visual summaries can be exported, and organising results by theme or method rather than paper by paper makes them clearer for readers.
  • The LibKey integration provides full-text access through institutional subscriptions, and Composio connects Google Drive so personal libraries can be analysed alongside new searches.
  • Every claim in the final report should remain traceable to a cited source so readers can check the evidence for themselves.
Apply it

To apply this, draft your review question in Deep Review Mode, gather results in a Combined Paper Table, and export a themed report once you have checked the key sources.

How to do it

  1. Frame questions with Deep Review Mode

    Write precise, researchable questions using Deep Review Mode in SciSpace Agent. A sharp question is the starting point for a literature review with AI agents.

  2. Search databases into a Combined Paper Table

    Search multiple databases at once and collect the results in a Combined Paper Table. This puts all your sources in one place.

  3. Filter and analyse findings interactively

    Filter, organise and analyse the findings to spot research gaps and emerging trends. Interactive analysis speeds up synthesis.

  4. Export citable reports

    Export structured, citable literature review reports with visual summaries. Verify the sources before you share the report.

  5. Use LibKey and Composio integrations

    Use LibKey for full-text access and Composio to analyse your Google Drive libraries. These integrations extend what the agent can read.

Watch the tutorial31 min

AI Literature Review: Discover, Analyse and Present with SciSpacePlay video: AI Literature Review: Discover, Analyse and Present with SciSpace

Chapters

  1. How to form a research question with SciSpace Deep Review
  2. Can AI generate research questions? SciSpace Agent demo
  3. How to search Google Scholar, PubMed and arXiv at once with Boolean queries
  4. Why run multiple queries per database and rerank the top 100 papers
  5. How to extract methods, results and limitations into table columns
  6. Bookmark papers to your SciSpace library and cite them
  7. How to find research gaps with AI using limitations columns
  8. How to prompt an AI research agent: name the columns and steps
  9. Extract insights: a citation-backed research gaps report
  10. Can AI write a literature review in LaTeX with APA, MLA or IEEE citations?
  11. Critical literature review with deep search in SciSpace Agent
  12. How the agent writes a two-column LaTeX review with a BibTeX file
  13. Types of literature reviews and how to do a scoping review with AI
  14. Charts and visualizations in an AI scoping review report
  15. Save the review PDF to your library and chat with it
  16. Systematic literature review with an AI agent: planning the SLR
  17. AI screening with inclusion and exclusion criteria
  18. PRISMA flow diagram and the final systematic review report
  19. Treat an AI research agent like a research assistant

Try this on SciSpace

Frequently asked questions

How do I write a good research question for a literature review?

A good review question is specific about the population or context, the concept or intervention, and the outcomes of interest, and it is narrow enough to answer with available studies. Frameworks such as PICO or SPIDER help structure it. Test the question with a quick search and refine it if results are too broad or sparse.

How do I present the results of a literature review?

Organise findings by theme, method or chronology rather than paper by paper, and use comparison tables or simple visuals to show patterns. Highlight agreements, contradictions and gaps, and connect them to your research question. Every claim should be traceable to a cited source so readers can check the evidence themselves.

Related tutorials