Quick Start

This guide covers every Elicit feature:
- Getting Started — Create your account and run your first search
- How to Use Notebook — Chat with papers and build a working research space
- How to Use Data Extractor — Pull key information out of PDFs into a table
- How to Use Paper Researcher — Turn a research question into a sourced report
- How to Use Topic Finder — Map a field before you commit to a research scope
- How to Use Custom Column — Extract any detail you define across every paper
- How to Use Research Alerts — Get told when additional papers match your topic
- How to Use Medtech — Screen clinical trials and medical evidence faster
- How to Use AI for Consumer — Run market and consumer research on published data
- How to Use Academic Research — Handle a full systematic review end to end
Time needed: 5 minutes per feature
Also in this guide: Pro Tips | Common Mistakes | Troubleshooting | Pricing | Alternatives
Why Trust This Guide
I’ve used Elicit for eight months across two literature review projects.
Every screen and setting below comes from my own work, not a vendor demo.

Elicit is an AI research assistant built on language models that automate research tasks.
It searches over 138 million academic papers using semantic understanding, not keyword matching.
The Elicit AI research assistant can find relevant papers and pull the data straight out of them.
It can automate research workflows that most people still run by hand.
Most users only ever run a search and stop there.
This guide shows you how to use Elicit properly, feature by feature.
Elicit Tutorial
This How to Use Elicit AI tutorial walks through every feature, from your first search to full systematic review workflows.

Elicit
Ask a research question in plain English and get the most relevant papers back in seconds. Elicit reads the PDFs, pulls out key findings with supporting quotes, and cuts systematic review time by up to 80%. Start free — no card needed.
Getting Started with Elicit
Before using any feature, finish this one-time setup.
It takes about three minutes.
Step 1: Create Your Account
Go to Elicit’s website and click Sign Up.
A free account gives you immediate access to search and basic extraction.
✓ Checkpoint: Check your inbox for the confirmation email.
Step 2: Open the Home Page
Elicit runs in the browser, so there is nothing to install.
Log in and you land on the home page with a single search bar.
Here’s what this looks like:

✓ Checkpoint: You should see the search bar and your empty library.
Step 3: Run Your First Search
Type a full research question in natural language instead of keywords.
Try something like: does sleep deprivation affect working memory in adults?
✅ Done: You’re ready to use any feature below.
How to Use Elicit Notebook
Notebook lets you chat with a set of papers and keep your notes beside the evidence.
Here’s how to use it step by step.
Step 1: Create a Notebook
Open Notebook from the sidebar and give your project a short description.
Step 2: Add Papers
Pull results from a search, or upload PDFs you already have on file.
Here’s what this looks like:

✓ Checkpoint: Your papers appear in a list with an AI summary beside each one.
Step 3: Ask Questions
Use the chat box to ask what the papers agree on and where they conflict.
✅ Result: A working research space where every answer links back to a source.
💡 Pro Tip: Keep one notebook per research question. Mixed topics make the chat vague and less useful.
How to Use Elicit Data Extractor
Data Extractor lets you pull key information out of PDFs into a clean, sortable table.
Here’s how to use it step by step.
Step 1: Upload PDFs
Drag your PDFs into the extractor, or select papers already saved in your library.
Step 2: Pick Your Fields
Choose sample size, methods and materials, outcome measures, or the conclusion.
Here’s what this looks like:

✓ Checkpoint: Each row is one paper and each column is one extracted field.
Step 3: Check the Quotes
Click any cell to see the supporting quotes Elicit pulled from the paper.
✅ Result: Extracting data from twenty papers now takes minutes instead of an afternoon.
💡 Pro Tip: Elicit hits 94-99% accuracy on empirical research, but always spot-check the numbers you plan to publish.
How to Use Elicit Paper Researcher
Paper Researcher lets you turn one research question into a sourced report you can actually cite.
Here’s how to use it step by step.
Step 1: Enter Your Question
Write the question as a sentence so the AI assistant understands intent, not keywords.
Step 2: Set the Scope
Tell it how deep to go — a quick answer or a full literature review.
Here’s what this looks like:

✓ Checkpoint: A structured report with inline references to every paper used.
Step 3: Read the Report
Elicit writes research reports that synthesise evidence across dozens of academic papers.
✅ Result: A 10+ page literature review draft built from real sources, not invented ones.
💡 Pro Tip: Ask follow-up questions inside the same report. Elicit keeps your earlier context and builds on it.
How to Use Elicit Topic Finder
Topic Finder lets you map a field visually before you commit to a narrow research question.
Here’s how to use it step by step.
Step 1: Enter a Broad Topic
Start wide. Type the general area you want to explore, not your final question.
Step 2: Browse the Clusters
Elicit groups related papers so you can see which sub-topics are crowded.
Here’s what this looks like:

✓ Checkpoint: A visual map of sub-topics with paper counts beside each one.
Step 3: Pick Your Angle
Choose a cluster with real evidence but few reviews, then narrow your search there.
✅ Result: A defensible research scope chosen from data instead of guesswork. You discover gaps early and focus your reading.
💡 Pro Tip: Use Topic Finder before your proposal. It surfaces gaps that a plain search will never show you.
How to Add Elicit Custom Column
Custom Column lets you extract any detail you define across every paper in your results.
Here’s how to use it step by step.
Step 1: Click Add Column
Open your results table and choose the option to create a new column.
Step 2: Write the Description
Describe exactly what to extract, such as funding source or dropout rate.
Here’s what this looks like:

✓ Checkpoint: A new column filled in across every paper in the table.
Step 3: Review Every Row
Scan the column and correct anything wrong — you can manually override any cell.
✅ Result: Custom fields no reference manager gives you, ready to analyze in one pass.
💡 Pro Tip: Phrase the column description as a question. Vague labels give vague answers.
How to Use Elicit Research Alerts
Research Alerts lets you get told when additional papers match a topic you already searched.
Here’s how to use it step by step.
Step 1: Save a Search
Run your search, then save it so Elicit remembers the exact query.
Step 2: Turn On Alerts
Enable the alert toggle and choose how often you want to hear about new work.
Here’s what this looks like:

✓ Checkpoint: A confirmation that the alert is active on that saved search.
Step 3: Triage the Email
Open each alert and send anything relevant straight to your Elicit library.
✅ Result: New papers reach you without you running the same search every week.
💡 Pro Tip: Set alerts on your two or three core questions only. More than that and you stop reading them.
How to Use Elicit Medtech
Medtech lets you screen clinical trials and medical evidence far faster than by hand.
Here’s how to use it step by step.
Step 1: Filter by Study Type
Restrict results to randomised trials, cohort studies, or meta-analyses.
Step 2: Add Screening Criteria
Write your inclusion and exclusion rules so Elicit can screen titles and abstracts.
Here’s what this looks like:

✓ Checkpoint: Screened papers marked include or exclude with a reason attached.
Step 3: Export the Set
Send the included papers to CSV, BibTeX, or RIS for your reference manager.
✅ Result: Automated screening that can cut up to 80% of manual screening time.
💡 Pro Tip: Always read the excluded pile once. A single bad criterion can drop the trial your review needs.
How to Use Elicit AI for Consumer
AI for Consumer lets you run market and consumer research against published academic evidence.
Here’s how to use it step by step.
Step 1: Frame the Business Question
Ask about behaviour, not products — pricing sensitivity or habit formation, for example.
Step 2: Filter by Date
Limit results to recent years so your insights reflect current consumer behaviour.
Here’s what this looks like:

✓ Checkpoint: A ranked set of studies with plain-language summaries beside each.
Step 3: Summarise the Findings
Have Elicit summarize the key findings into a short brief for your team.
✅ Result: Claims backed by peer-reviewed evidence instead of a competitor blog post.
💡 Pro Tip: Search the underlying behaviour rather than your industry. The best evidence usually sits in another field.
How to Use Elicit Academic Researcher
Academic Research lets you run a full systematic review from first search to final data table.
Here’s how to use it step by step.
Step 1: Start the Workflow
Open the Systematic Review workflow and enter your protocol question.
Step 2: Screen and Extract
Work through screening, then extract your outcome data into columns.
Here’s what this looks like:

✓ Checkpoint: A step-by-step checklist showing which stage you are on.
Step 3: Export Everything
Download the table and references, then write your synthesis around them.
✅ Result: A systematic review that took weeks now takes days, with the audit trail intact.
💡 Pro Tip: Elicit can save researchers up to five hours a week here. Log your search dates as you go.
Elicit Pro Tips and Shortcuts
After eight months of using Elicit daily, these are the habits that save time on every project.
The first one is simple: let it summarize long articles before you decide to read them.
Keyboard Shortcuts
| Action | Shortcut |
|---|---|
| New search | Ctrl / Cmd + K |
| Save paper to library | S |
| Open chat panel | Ctrl / Cmd + J |
| Close a paper view | Esc |
Hidden Features Most People Miss
- Load more results: Find Papers shows a first page only. Keep loading to reach 100 papers.
- Elicit’s content sources: Search draws on Semantic Scholar, so open access coverage is wider than you expect.
- Citation count filters: Filter by publication date, study type, and citations to tighten a messy result set.
- Export to RIS: Push references into Zotero or EndNote instead of retyping them by hand.
One more thing worth knowing about how Elicit works.
It ranks by meaning, so it surfaces relevant papers that are not a perfect keyword match.
Here’s my own workflow in practice:

I start broad, filter hard, then extract only the fields my write-up actually needs.
Elicit Common Mistakes to Avoid
Mistake #1: Searching With Keywords
❌ Wrong: Typing “sleep memory adults” like it is a database search.
✅ Right: Ask a full research question in natural language and let the AI research assistant read intent.
Mistake #2: Trusting Summaries Without Checking
❌ Wrong: Quoting an AI-generated summary straight into your own work.
✅ Right: Open the supporting quotes and verify every claim against the original paper.
Mistake #3: Extracting Everything
❌ Wrong: Adding fifteen columns because the tool makes extracting data easy.
✅ Right: Extract only what your analysis needs. Fewer columns means fewer errors to check.
Elicit Troubleshooting
Problem: Search Returns Off-Topic Papers
Cause: Your question is too broad or mixes two separate ideas.
Fix: Split it into two questions and add a date filter to narrow the scope.
Problem: A Column Comes Back Empty
Cause: The papers genuinely do not report that field, or your description was ambiguous.
Fix: Rewrite the column as a direct question, then manually override the rows you can fill yourself.
Problem: A PDF Will Not Process
Cause: Scanned PDFs without a text layer cannot be read by any AI assistant.
Fix: Run the file through OCR first, then upload it again.
📌 Note: If none of these fix your issue, contact Elicit support.
What is Elicit?
Elicit is an AI research assistant that automates the slow parts of the research process.
Think of it like a research assistant who has already read every paper in your field.
Watch this quick overview:
The video walks through the interface before we get into the detail.
Elicit uses artificial intelligence to find papers, rank them, and pull data out of them.
It includes these key features:
- Notebook: Chat with a set of papers and keep your notes beside the evidence
- Data Extractor: Pull key information out of PDFs into a clean, sortable table
- Paper Researcher: Turn one research question into a sourced report you can actually cite
- Topic Finder: Map a field visually before you commit to a narrow research question
- Custom Column: Extract any detail you define across every paper in your results
- Research Alerts: Get told when additional papers match a topic you already searched
- Medtech: Screen clinical trials and medical evidence far faster than by hand
- AI for Consumer: Run market and consumer research against published academic evidence
- Academic Research: Run a full systematic review from first search to final data table
For a full review, see our Elicit review.

In 2026 it remains the default tool for automating systematic review work and data extraction.
Elicit Pricing
Here’s what Elicit costs in 2026:
| Plan | Price | Best For |
|---|---|---|
| Basic (Free) | $0 | Trying search and light extraction |
| Basic | $9 | Students running one review at a time |
| Standard | $12 | Regular literature review work |
| Pro | $19 | Heavy extraction and research reports |
| Enterprise | Contact Sales | Labs and teams needing shared access |
Free trial: Yes — the free plan runs indefinitely with monthly credit limits.
Money-back guarantee: Not advertised, but paid plans cancel at any time.

💰 Best Value: Pro — the extraction credits are what make the tool pay for itself.
Elicit vs Alternatives
How does Elicit compare? Here’s the competitive landscape:
| Tool | Best For | Price | Rating |
|---|---|---|---|
| Elicit | Data extraction at scale | $9/mo | ⭐ 4.3 |
| Consensus | Fast yes/no answers | $8.99/mo | ⭐ 4.4 |
| Scite | Citation context | $20/mo | ⭐ 4.2 |
| SciSpace | Reading dense PDFs | $12/mo | ⭐ 4.1 |
| Research Rabbit | Citation mapping | Free | ⭐ 4.3 |
| Semantic Scholar | Open paper search | Free | ⭐ 4.4 |
| Connected Papers | Visual field maps | $6/mo | ⭐ 4.0 |
| Perplexity | General research chat | $20/mo | ⭐ 4.5 |
| Litmaps | Tracking new work | $10/mo | ⭐ 4.0 |
Most research tools stop at search, which is where the real comparison starts.
Quick picks:
- Best overall: Elicit — nothing else extracts structured data across papers this well
- Best budget: Semantic Scholar — free access to the same underlying paper index
- Best for beginners: Consensus — one question, one answer, no setup
- Best for systematic review: Elicit — the screening workflow is built for it
🎯 Elicit Alternatives
Looking for Elicit alternatives? Here are the top options:
- 🚀 Consensus: Answers a research question in one line, then shows the studies behind it. Faster than Elicit for quick checks.
- 💰 Semantic Scholar: Free academic search across the same corpus Elicit draws on. No extraction, but excellent coverage.
- 🎨 Connected Papers: Builds a visual graph of related work from one seed paper. Good for finding what you missed.
- ⚡ SciSpace: Explains dense passages line by line while you read. Strong on making hard papers make sense.
- 🔒 Scite: Shows whether later work supports or disputes a claim. Useful for judging how solid the evidence is.
- 🧠 Perplexity: General AI assistant with academic search built in. Broader scope, less depth on empirical research.
- 👶 Research Rabbit: Free citation explorer with a gentle learning curve. Great first tool for new researchers.
- 📊 Litmaps: Tracks a topic over time and mails you additional papers. Similar idea to research alerts.
For the full list, see our Elicit alternatives guide.
⚔️ Elicit Compared
Here’s how Elicit stacks up against each competitor:
- Elicit vs Consensus: Consensus wins on speed. Elicit wins the moment you need a table of extracted data.
- Elicit vs Semantic Scholar: Semantic Scholar is free search only. Elicit reads the papers and summarises them for you.
- Elicit vs Connected Papers: Connected Papers maps relationships. Elicit answers the question and extracts key findings.
- Elicit vs SciSpace: SciSpace is better for reading one paper. Elicit is better across a hundred.
- Elicit vs Scite: Scite judges claim reliability. Elicit finds the papers and builds research reports around them.
- Elicit vs Perplexity: Perplexity covers everything. Elicit stays on academic papers and cites them properly.
- Elicit vs Research Rabbit: Research Rabbit is free and visual. Elicit adds extraction and screening that Rabbit lacks.
- Elicit vs Litmaps: Litmaps monitors a field. Elicit does that plus the full systematic review workflow.
Start Using Elicit Now
You learned how to use every major Elicit feature:
- ✅ Notebook
- ✅ Data Extractor
- ✅ Paper Researcher
- ✅ Topic Finder
- ✅ Custom Column
- ✅ Research Alerts
- ✅ Medtech
- ✅ AI for Consumer
- ✅ Academic Research
Next step: Pick one feature and try it now.
Most researchers start with Paper Researcher.
It takes less than five minutes to see whether the tool fits your work.
Frequently Asked Questions
How do you use Elicit?
Create a free account, type your research question in natural language, then review the ranked papers. Add columns to extract key information, or open Notebook to chat with the results.
Can I use Elicit for free?
Yes. The Basic (Free) plan costs $0 and includes search and limited extraction credits. Paid plans start at $9 and add higher limits for heavier literature review work.
How does Elicit generate results?
Elicit uses language models and semantic understanding to search over 138 million academic papers. It ranks by meaning, summarises each paper, and shows supporting quotes for every extracted field.
Is Elicit good for research?
Yes, especially for empirical research. It reaches 94-99% accuracy on data extraction and can cut systematic review time by up to 80% when you verify its output.
How is Elicit different from ChatGPT?
Elicit searches a real academic database and cites the papers it used. ChatGPT generates from training data and can invent references, so it is riskier for citation work.













