Paperguide logo
Academic Research · Rag Semantic Search

Paperguide

Paperguide is an AI research platform that unifies literature search, reference management, data extraction and cited writing for researchers and evidence synthesis teams, with every claim traceable to its source sentence across 200M+ peer-reviewed papers.

Active Free plan Freemium API available Verified by Guidaio
Overview

What is Paperguide?

Paperguide is an AI-native platform for scientific research workflows and evidence synthesis, published by Keevs Inc. It is built as four connected blocks rather than a single feature: AI Search for discovery, a Reference Manager for organisation, a Research Agent paired with Extract Data for synthesis, and an AI Writer for drafting.

Discovery runs on a corpus of more than 200 million peer-reviewed papers aggregated from PubMed, arXiv, OpenAlex and Semantic Scholar. AI Search reads a natural-language question, re-ranks results on relevance and journal quality signals — SJR quartile, SNIP, SJR score, citation metrics — and synthesises an answer from the twenty strongest papers. The Reference Manager imports Zotero and Mendeley libraries directly, EndNote through BibTeX or RIS, and accepts papers by DOI, URL or PDF, pulling metadata and open-access PDFs automatically. A built-in reader carries highlights, annotations and notes with the paper across projects.

The systematic review module is what sets the platform apart. A systematic review lives or dies by its method: a predefined protocol, a documented search, recorded screening decisions and a matching PRISMA flow generated from the review's real counts. Two modes exist. In Dual Review, two reviewers screen independently with a conflict resolver, and the AI never decides — it prepares the cited evidence. In AI Systematic Review, the AI rules and a named verifier confirms each stage. Either way the AI surfaces the relevant statements for each criterion and cites them in place, so a reviewer reads a few sourced sentences instead of a whole paper.

Extract Data turns selected papers into a table whose columns you define with an instruction, choosing between a free-text Answer, Yes/No or a fixed option list; every value links back to the sentence it came from, and the table exports to CSV or Excel. The Research Agent handles Chat with PDF, multi-step investigation and a Literature Review Report over up to 50 papers organised by theme. An MCP server plugs the corpus into Claude, ChatGPT or Cursor, and an institutional proxy such as EZproxy unlocks paywalled articles.

Shared workspaces with Edit, Reviewer and View-only roles serve the 974,000+ researchers Paperguide claims, across medical devices, HEOR consulting, pharma and life sciences, government and academia. The stated limit: it runs in the browser, with no offline access and no mobile app.

What it does

  • Ask a research question in plain language and get a synthesised, cited answer drawn from 200M+ peer-reviewed papers
  • Import your Zotero or Mendeley library, or add papers one by one by DOI, URL or PDF
  • Run a systematic review with a predefined protocol, independent double screening and a generated PRISMA flow diagram
  • Extract data into a table whose columns you define, with every value linked back to its source sentence
  • Draft a cited document from your own library using more than 1,000 citation styles
  • Connect the MCP server to Claude, ChatGPT or Cursor and search without leaving your assistant
  • Export your work as RIS, CSV, BIB, PDF, DOCX or Excel
Audience

When to use Paperguide / When not to

A quick filter to help you decide if Paperguide is the right fit.

When to use Paperguide

  • Researchers and PhD candidates running literature reviews who would rather have one workspace than three to five separate tools
  • Evidence synthesis teams producing PRISMA systematic reviews for a journal, an HTA dossier, a clinical evaluation report or a regulatory audit
  • Regulatory affairs and medical device professionals who must defend every exclusion and every extracted figure in front of a reviewer or an auditor
  • HEOR consultants and pharma or life sciences teams screening large volumes of papers under deadline; one customer screened and extracted a 2,000-paper PRISMA review in under a week
  • Government and public policy analysts who need synthesised, citable answers drawn from peer-reviewed literature rather than open-web summaries

When not to use Paperguide

  • Anyone who needs to work offline: Paperguide runs in the browser today, so an internet connection is required
  • Mobile-first users, since there is no iOS or Android application
  • Organisations that require EU data residency, as all data is hosted in the United States on Microsoft Azure
  • Buyers whose procurement demands GDPR, SOC 2 or ISO 27001 certification, all three of which are still announced as coming soon
  • Teams working on internal, non-academic corpora rather than published scientific literature
Get started

How to use Paperguide

A typical end-to-end flow, from setup to results.

  1. Create a free account from Start for free — no card is requested on the Free plan
  2. Import your existing library from Zotero or Mendeley in a single step, or from EndNote via BibTeX or RIS
  3. Add papers by DOI, URL or PDF upload, or capture them from any web page with the Chrome extension
  4. Enter your institutional proxy URL (EZproxy) to retrieve the paywalled articles your institution licenses
  5. Ask a question in plain language in AI Search, then save the cited results into your library
  6. For a systematic review, build the protocol first: the AI proposes eligibility criteria and extraction fields, you edit them, and the order of the criteria determines the exclusion reasons
  7. Let search results and library items be pooled and deduplicated before screening begins, and route any paper found along the way through the same screening path so the PRISMA counts stay honest
  8. Assign roles stage by stage — reviewers, verifiers and conflict resolver
  9. For extraction, define each column with a title and an instruction, choose its output type, run it on the selected papers, then export the table to CSV or Excel; RIS, BIB, PDF and DOCX are available at other stages
  10. Connect the MCP server if you work inside an assistant: Settings > Connectors > Add custom connector in Claude, paste the server URL leaving client ID and secret empty, then sign in with your Paperguide account
Quick read

Pros & Cons

Pros

  • Sentence-level traceability on every claim, every answer and every extracted value
  • A genuinely tooled systematic review: predefined protocol, independent double screening, conflict resolver and a PRISMA diagram built from the review's real counts
  • The final decision stays human and recorded; in Dual Review mode the AI never rules, it only prepares the cited evidence
  • A broad, open corpus — PubMed, arXiv, OpenAlex, Semantic Scholar — with source quality signals (SJR, SNIP, citations) visible before you cite
  • Replaces several tools with a single workspace; customers report going from five tools to one, and three separate tools becoming one workspace
  • Low-risk entry: a permanent free plan with 2,000 AI credits a month, MCP included and 10 API requests, then 9 USD per seat and per month on annual commitment, with a 30-day money-back guarantee on the first month
  • Ownership of your content and of the AI outputs is stated without ambiguity in the terms, and the site interface is available in seven languages

Cons

  • No certification held: GDPR, SOC 2 and ISO 27001 are all announced as coming soon
  • A generic online-store privacy policy that talks about orders, shipping addresses and credit cards, poorly aligned with a research SaaS
  • No published DPA, no subprocessor list, no Article 27 EU representative and no named DPO, with a single address — hello@paperguide.ai — for support, legal and GDPR alike
  • Hosting in the United States only (Microsoft Azure), with transfers outside the EU explicitly announced to Canada and the United States
  • The no-training clause is limited to public or generalized AI models, and the reservation that wording leaves is never explained
  • Prices are displayed on annual commitment by default and the monthly rates are not rendered without JavaScript, while plans described as unlimited are subject to a fair-use policy with rate limiting
  • No offline access and no mobile app, and several headline features are still marked coming soon (image and table extraction, PRISMA dual reviewer screening on Evidence)
Pricing

Pricing & Plans

Paperguide operates on a freemium model. A permanent Free plan is available at no cost, alongside four paid tiers, one of which is quoted on request. The lowest paid entry point is 9.00 USD per seat and per month on the Plus plan, billed annually at 108 USD; the pricing page displays annual commitment rates by default and the corresponding monthly rates are not rendered in static HTML. Purchases are covered by a 100% money-back guarantee within 30 days of the initial purchase, applicable to the first month of the subscription only. A subscription may be cancelled at any time, with access retained until the end of the current billing cycle. AI credits act as the common currency across Research Agents, Research Reports, AI Writer and Systematic Reviews, and the MCP server draws on those same credits without a separate connector or API charge. A fair-use policy applies rate limiting to the plans described as unlimited.

Free — For casual exploration. No cost, 1 seat
  • 2
  • 000 AI credits a month
  • Research Agent
  • limited Reference Manager
  • limited AI searches
  • AI Writer
  • up to 5 columns and 10 papers per extraction table
  • Literature Review Report
Pro — For power researchers, flagged Popular. 29 USD per seat and per month on annual commitment (348 USD a year, advertised as Save 41%)
  • 50
  • 000 AI credits a month
  • systematic reviews up to 5
  • 000 papers
  • image and table extraction announced as coming soon
  • everything else as in Plus
  • and 500 Search API requests a month
Evidence — For systematic reviews, flagged Teams. 69 USD per seat and per month on annual commitment (828 USD a year, advertised as Save 30%)
  • 150
  • 000 AI credits a month
  • systematic reviews up to 5
  • 000 papers
  • PRISMA-grade dual reviewer screening announced as coming soon
  • a dedicated support channel and 1
  • 500 Search API requests a month
Enterprise — For companies & institutions. Custom pricing, through the sales team
  • custom AI credits
  • systematic reviews up to 20
  • 000 papers
  • custom columns and papers per extraction table
  • and custom Search API requests
Special offers — A Faculty & Student discount is announced, through an external page hosted on Notion; the site does not publish the discount rate · Annual billing is advertised as saving up to 43% — Save 40% on Plus, Save 41% on Pro and Save 30% on Evidence · A 100% money-back guarantee within 30 days of the initial purchase, applicable to the first month of the subscription only · An Industry / Academic toggle sits on the pricing page, though the academic grids are not rendered in static HTML
Prices and plans listed above may evolve. Always check the official pricing page before subscribing.
Trust & Privacy

Data, GDPR & hosting

A consolidated view of how Paperguide handles your data.

GDPR overview

Paperguide claims no GDPR compliance today. The home page lists GDPR, SOC 2 and ISO 27001 as coming soon, which is a roadmap statement rather than an attestation. The privacy policy does grant European residents the right to access the personal information held about them and to ask that it be corrected, updated or deleted, exercised by email, and it points to contract performance and legitimate interests as legal bases. It also states plainly that information will be transferred outside Europe, including to Canada and the United States, and that the site does not alter its data collection practices when it receives a Do Not Track signal. What is missing is substantial: no published DPA, no subprocessor list, no Article 27 EU representative, no named DPO, and no address or entity in the European Union. The only contact point is hello@paperguide.ai.

Who owns the data?

Paperguide's terms are unusually direct on ownership. Section 2A.1 states that you retain full ownership of every document, file, prompt and dataset you upload, and that Paperguide claims no ownership rights in that User Content. Section 2A.4 extends the same rule to machine output: AI-generated summaries, analyses and citations created from your content belong to you, and the publisher claims no right to reuse or redistribute them. What Paperguide takes instead is a limited, non-exclusive, non-transferable and revocable licence under section 2A.2, granted solely to operate the service. One carve-out is worth knowing: product feedback and suggestions you send fall outside this regime under section 9 and can be used without compensation.

Reuse rights

Because ownership of both your uploads and the AI outputs stays with you, you can reuse, publish, export and redistribute what you extract or draft without asking permission. On the publisher's side, section 2A.3 says Paperguide treats User Content as confidential: it does not sell your data, does not share it with third parties for commercial purposes, and does not use it to train public or generalized AI models. Read literally, that last phrase is a commitment about a category of models rather than a blanket no-training pledge, and the reservation it leaves is never spelled out. Section 2A.5 allows aggregated, anonymised usage data — data that identifies neither you nor your content — to be used for service improvement and reliability. The licence granted in section 2A.2 explicitly excludes resale, disclosure and marketing use. On the public website the picture is looser: the privacy policy mentions Google Analytics and targeted advertising through Facebook, Google and Bing, and reads like a generic online store policy rather than one written for a research platform.

Data retention & training

Retention summary
Paperguide publishes rules but no durations. Section 2A.6 of the terms states that on termination, User Content will be deleted or anonymized according to retention policies, and that confidentiality obligations survive termination — without saying what those retention policies contain or how long anything is kept. The privacy policy's data retention section is thinner still and store-shaped: order information is kept for the company's records unless and until you ask for it to be deleted. European residents are told they may exercise a right of erasure, by email. No figure, no detailed retention schedule and no publicly accessible retention policy exist, so the practical answer is that deletion happens on request or on termination, on a timeline the publisher does not commit to.
Trains on customer data
No
GDPR contact

Hosting summary

Paperguide states in its pricing FAQ that all Paperguide data is stored in a US location, on Microsoft Azure servers, protected by encryption and secure servers, and that data is neither shared nor sold. The references FAQ adds that research data is encrypted and stored securely, with regular backups. No EU region is offered and there is no data residency option, so a European institution has no way to keep its data inside the Union. The privacy policy confirms the direction of travel, announcing that information will be transferred outside Europe, including to Canada and the United States. One technical caveat is worth stating plainly: the site's DNS resolves to 76.76.21.21, an anycast node on AS16509 (Amazon.com Inc.). That describes where the marketing site is served, not necessarily where application data lives, and it should not be read as contradicting the Azure statement. Beyond the jurisdiction and the provider name, no further hosting detail is published.

Hosting countries
🇺🇸 United States
Hosting regions
North America
Availability

Where Paperguide works

Country-level availability.

Not available in

No country or region is announced as unavailable. Section 5 of the terms reserves the right to limit sales to any person, geographic region or jurisdiction on a case By Case basis, but this is a generic reservation clause rather than a list of excluded countries.
Watch-outs

Things to keep in mind

Risks and trade-offs to weigh before adopting Paperguide.

  • GDPR, SOC 2 and ISO 27001 are announced as coming soon: do not treat them as acquired when you assemble a procurement or ethics file
  • The privacy policy is generic and speaks of orders, shipping and credit cards; it does not describe how a research platform actually processes your data
  • The no-training clause targets public or generalized AI models only, and the wording leaves a reservation that is never explained
  • All data is hosted in the United States, transfers outside the EU to Canada and the United States are announced, and no DPA, subprocessor list or Article 27 EU representative is published — with a single channel handling support, legal and GDPR requests
  • Prices are shown on annual commitment by default: the real monthly cost is higher and is not visible without JavaScript, and plans described as unlimited are subject to fair-use rate limiting
  • Several features advertised on the plan pages are marked coming soon; check what actually ships before buying for a specific need
  • Cited sentences are easy to accept at face value, and reading a few sourced statements instead of the full paper can quietly erode the critical reading a review depends on: a systematic review remains a human responsibility, and auditability rests on your team's discipline rather than on the software
Setup

Setup & Integrations

Technical difficulty

Low for everyday use. Paperguide is a web tool with nothing to install: a free account takes a few clicks and a Zotero or Mendeley library imports in one step. Two optional pieces raise the bar. The Chrome extension must be installed for web capture and institutional library access, and the institutional proxy (EZproxy) needs a URL entered, usually an administrator or librarian task. Connecting the MCP server is a stated three-step operation. No development skills are needed for normal use; only the Search API assumes technical ability. On a systematic review, the load is methodological rather than technical.

Deployment

Web appAPIBrowser extensionChrome extension

Integrations

Zotero Mendeley EndNote EZproxy PubMed ArXiv OpenAlex Semantic Scholar Claude ChatGPT Cursor Microsoft Word Microsoft Excel Google Chrome

Supported languages

EnglishSpanishFrenchGermanItalianJapaneseChinese
Company

Behind Paperguide

Company name
Keevs Inc
Founded
27/07/2024
Country of origin
🇺🇸 United States
Headquarters
2925 Manor Bridge Dr, Alpharetta, GA 30004, United States
US office
2925 Manor Bridge Dr, Alpharetta, GA 30004, United States
UBO
INFORMATION_NOT_FOUND
UBO country
INFORMATION_NOT_FOUND
Domain registrar country
🇺🇸 United States
Legal contact
Support contact

Social

Official links

Resources

All the official URLs gathered for verification and reference.

Compare

Alternatives

Tools that compete with or complement Paperguide.

P PaperpileS ScispaceZ ZoteroM MendeleyG Google ScholarJ Jenni AIA AfforaiC Consensus
FAQ

Frequently asked questions

Is there a free plan?
Yes. Paperguide keeps a permanent Free plan with 2,000 AI credits a month on one seat, and no card is requested to open it. It includes the Research Agent, AI Writer, Chat with PDF, Literature Review Report, Zotero and Mendeley import, the Paperguide MCP server and 10 Search API requests a month, with the Reference Manager, AI searches and extraction tables capped.
Can I cancel at any time, and is there a refund?
You can cancel whenever you want and keep access until the end of the current billing cycle. Paperguide also offers a 100% money-back guarantee within 30 days of the initial purchase, applicable only to the first month of the subscription.
Where is my data hosted?
In the United States, on Microsoft Azure servers. Paperguide states that data is protected by encryption and secure servers, backed up regularly, and neither shared nor sold. No EU region or data residency option is offered, and the privacy policy announces transfers outside Europe, including to Canada and the United States.
Is my content used to train AI models?
The terms say no for a defined category: section 2A.3 states that Paperguide does not use User Content to train public or generalized AI models, and treats that content as confidential. The wording is about those model types specifically rather than a blanket no-training commitment, and the reservation is not explained. Aggregated, anonymised usage data may be used for service improvement under section 2A.5.
Can I import my library from Zotero, Mendeley or EndNote?
Yes. Zotero and Mendeley import directly in one step, and EndNote comes across via BibTeX or RIS. You can also add papers by DOI, URL or PDF upload, and Paperguide retrieves metadata and open-access PDFs automatically.
How do I reach paywalled articles?
Through your institutional proxy. Paperguide supports EZproxy and other major providers: you enter your proxy URL and the platform retrieves the paywalled papers your institution licenses. The Chrome extension also opens institutional library access from any web page.
Does the AI make screening decisions for me in a systematic review?
It depends on the mode. In Dual Review, two reviewers screen independently with a conflict resolver and the AI never decides — it prepares the cited evidence for each criterion. In AI Systematic Review, the AI rules but a named verifier confirms each stage. In both cases the final decision is human and recorded, and the PRISMA diagram is generated from the review's real counts.
Can I plug Paperguide into Claude or ChatGPT?
Yes, through the Paperguide MCP server, which also works with Cursor. There is no connector fee and no separate API plan: it consumes the same AI credits as the rest of the application. Separately, a Search API is available with a monthly quota per plan, from 10 requests on Free up to 1,500 on Evidence, and custom volumes on Enterprise.
Is Paperguide GDPR compliant?
Not as things stand. The home page announces GDPR, SOC 2 and ISO 27001 as coming soon, so none of them is held today. European residents are granted access, rectification and erasure rights by email, but no DPA, subprocessor list, Article 27 EU representative or named DPO is published.
Is there offline access or a mobile app?
No to both. Paperguide runs in the browser today, so an internet connection is required, and there is no iOS or Android application. Access goes through the web app, the Chrome extension, the API or the MCP server.
Conclusion

Should you pick Paperguide?

Paperguide is a solid, deliberately specialised tool. It sits on open, well-regarded corpora — PubMed, arXiv, OpenAlex, Semantic Scholar — and, unlike most AI research assistants, it treats the systematic review as a method rather than a feature: a predefined protocol, a documented search, recorded screening decisions and a PRISMA diagram built from the review's real counts.

Its real differentiator is not the size of the index but the discipline around it. Every claim, every answer and every extracted value points back to the sentence it came from, and the final call stays human and recorded — in Dual Review mode the AI never decides, it only prepares the cited evidence. For anyone who will have to defend an exclusion to a reviewer or an auditor, that is the difference between a shortcut and a usable method.

The entry point carries little risk. The free plan is permanent, Zotero and Mendeley libraries import in a single step, and the first paid tier costs 9 USD per seat and per month on annual commitment, with a 30-day money-back guarantee covering the first month.

The reservation is compliance maturity. Nothing is certified: GDPR, SOC 2 and ISO 27001 are all announced as coming soon. The privacy policy reads like a generic online store, with orders, shipping and credit cards, and does not describe how a research platform actually handles data. Hosting is United States only, transfers outside Europe are explicit, and a single address handles support, legal and GDPR requests alike.

The verdict is therefore split. For academic research and evidence synthesis, Paperguide is convincing and, at this price, easy to try. For a regulated European setting, treat the compliance file as unfinished business to investigate before adoption rather than a box already ticked.