Choose one of three approaches. The steps for that option appear below.
Choose an option above. On a phone, swipe the tabs left or right to see all options →
Structured gather, screen, extract and report steps; the Systematic Review workflow needs a Pro, Scale or Enterprise plan.
Get studies and question ready
Screen, extract into columns
Check decisions against texts
Generate the draft report
Lead, reviewer check, decide
Use: You
Have ready: the bounded question, the selected full-text studies and the search limits; the permitted-use limits and any supplied decision criteria.
Check every study is full text and the question statement matches the agreed scope.
Use: Elicit
Upload the studies with the scope statement; screen for inclusion or exclusion, then extract findings with one column per data point: population, method, finding, limitation.
What to paste into the tool
Task: Screening decisions with reasons; Extraction table with cited cells. Screen the uploaded studies against the stated question and give an inclusion or exclusion reason for each. For the included studies add columns for population, method, main finding with exact figures and stated limitations, citing the study text in each cell. Use only the uploaded studies.
Use: You
Check each screening decision and extracted cell against the full texts, confirm scope and search limits, and record unresolved fields in the evidence ledger.
Check every ledger cell was compared with the full text of its study.
Use: Elicit
Generate the PRISMA-style draft report from the checked ledger, organised by agreement and disagreement within the bounded scope, stating the coverage limitations.
What to paste into the tool
Task: Report draft with findings and coverage limitations; Source-linked uncertainty and exception notes. Write the report from the screened and extracted studies only. Group findings by agreement and disagreement, and list the coverage limitations: the bounded question, the search limits and any assessment step that was omitted. Add nothing that is not in the ledger.
Use: The research lead and qualified reviewer
Check conclusions against the extraction table; route methodological or statistical interpretation to the qualified reviewer; record corrections and unresolved items; no exhaustive-coverage claim.
Consensus searches its own database and synthesises with citations; the analyst reconciles coverage against the supplied studies and people decide.
Word the question neutrally
Search, collect the synthesis
Reconcile papers, check passages
Follow up, draft review
Lead, reviewer check, decide
Use: You
Have ready: the bounded question, the selected studies and the search limits. Rewrite the question and search limits as neutral search wording.
Check the wording matches the bounded question and its limits.
Use: Consensus
Search the question in Consensus; collect the synthesis with citations, the papers used, and the stated agreement and disagreement.
What to paste into the tool
Task: Synthesis with citations; List of papers used; Agreement and disagreement summary. For this bounded question, summarize what the research says with citations to the papers used. Say where studies agree and where they disagree. If the evidence found is not enough, say so plainly. Search and summarize only.
Use: You
Reconcile the paper list with the supplied studies, open each cited passage and confirm it supports its claim; record which supplied studies were covered.
Check every cited passage was opened and supports the claim attached to it.
Use: Consensus
Run a follow-up search per disagreement point for contrasting papers and passages; draft the review within the bounded scope, naming coverage limitations.
What to paste into the tool
Task: Contrasting papers with cited passages. For each of these disagreement questions find the specific papers on each side and cite the passage that states each finding. If the evidence found is thin, say so. Search and cite only. Task: Review draft. From this checked synthesis and these contrasting papers, draft a short review within the bounded scope: the findings, where studies agree and disagree, and the coverage limitations. Keep the citation on every finding. If the evidence is thin, say so plainly. Search and summarize only.
Use: The research lead and qualified reviewer
Check conclusions against the checked passages; route methodological or statistical interpretation to the qualified reviewer; record corrections and unresolved items; no exhaustive-coverage claim.
The review stays within the supplied studies; the notebook answers with quotes and drafts only from checked notes.
Get studies and scope ready
Findings, then a grouping
Check quotes, confirm grouping
Draft the review
Lead, reviewer check, decide
Use: You
Have ready: the bounded question, the selected full-text studies and the search limits; the permitted-use limits and any supplied decision criteria.
Check only the selected studies are included and the scope text matches the agreed question.
Use: Gemini Notebook
Create a notebook from the studies; ask finding-by-finding questions per study with quote citations, then draft a grouping by agreement and disagreement plus coverage limitations.
What to paste into the tool
Task: Per-study findings with quote citations. For each uploaded study give its main finding with exact figures, its method, its population and its stated limitations. Quote the passage for each. If a study does not state one of these, write not stated. Use only the uploaded studies. Task: Draft grouping by agreement and disagreement; Draft coverage limitations list. Using only the cited findings in this notebook, group the findings by agreement and disagreement with one citation per finding, and list the coverage limitations: the bounded question, the search limits and any assessment step that was omitted. Add nothing that is not in the cited findings.
Use: You
Check each quoted finding at its place in the full text, confirm the drafted grouping and coverage limitations, and write the checked notes.
Check each grouped finding against its full text before the notes are used.
Use: Gemini Notebook
Add the checked notes to the notebook; ask it to draft the review from them, with one citation per finding and an explicit limitations section.
What to paste into the tool
Task: Notebook review draft; Source-linked uncertainty and exception notes. Using only the checked notes in this notebook, draft the review: findings grouped by agreement and disagreement, one citation per finding, and a limitations section naming the bounded question, the search limits and the omitted assessment steps. Add nothing that is not in the notes.
Use: The research lead and qualified reviewer
Check conclusions against the checked passages; route methodological or statistical interpretation to the qualified reviewer; record corrections and unresolved items; no exhaustive-coverage claim.
Best when: One workflow for screening, extraction and a draft report
What it does: Elicit screens and extracts; people check and correct
Strong at: One column per data point to extract
Limit: The Systematic Review workflow needs a paid plan
Free plan: Yes, limited (the Systematic Review workflow needs a paid plan)
Best when: A scan beyond the supplied studies
What it does: Consensus searches and synthesises; you reconcile coverage
Strong at: Searches over 220 million peer-reviewed papers
Limit: Supplied studies may be missing from its database
Free plan: Yes, limited (10 Pro messages, 3 Deep reviews a month)
Best when: A review confined to the supplied studies
What it does: Gemini Notebook quotes findings; you check, it drafts
Strong at: Drafts only from notes a person checked
Limit: Coverage is only as wide as the supplied set
Free plan: Yes, limited (50 sources per notebook)
Before you paste into any AI tool, remove any confidential information. Put it back at the end.
Never paste credentials or full payment details.
The research analyst checks screening decisions, extracted passages and findings against the full texts before the review is written.
The subject research lead checks conclusions and blocks any claim of exhaustive coverage.
Methodological or statistical interpretation goes to the named qualified reviewer, and if the studies carry medical or other high-stakes implications the relevant qualified professional reviews before reuse.
Check the result against your original inputs. Approve it before sharing or acting.