ResearchCollab
ResearchCollab is an AI research assistant that searches over 460 million papers, maps a question into an interactive mind map, runs PRISMA-aligned systematic reviews, and drafts fully cited sections from your own library.
What is ResearchCollab?
ResearchCollab is a London-based web platform that puts an AI layer over academic literature, built around a claimed index of more than 460 million papers. It organises the work into three stages the site calls Find, Review and Write.
Find begins with a plain-language question. The Topic Map turns it into an interactive mind map of related papers, sub-topics and the gaps worth chasing, with paper counts per area so you can gauge how much evidence already exists. Semantic search means you do not need perfect keywords.
Review is where the product makes its strongest claim. A systematic literature review runs across seven stages mapped onto the four PRISMA phases — identification, screening, eligibility and inclusion — with inclusion and exclusion tracked as you go and the PRISMA flow diagram generated for you, ready to drop into a methods section. An Evidence Tracker keeps every claim tied to the paragraph it came from.
Write covers the drafting end: an Outline Generator, section drafting from your Knowledge Bank, and Virtual Peer Review, a critique engine meant to give you a fresh read before a supervisor does. Anyone who prefers to delegate can hand the whole chain to Research Agent, which reads the question, searches, filters for relevance and returns a cited note.
Traceability is the pitch against general assistants: every claim links back to a paragraph in a real indexed paper, and the site explicitly rejects fabricated DOIs. Uploaded documents are stated not to be used for model training.
Around this sit a Knowledge Bank for findings and annotations, shared workspaces with role-based permissions, full-text search across your library, reusable prompt templates, and a browser extension for saving scholarly material from the web. Plus and Professional unlock 50+ models including GPT, Gemini, Claude, Perplexity, Grok and DeepSeek, plus Deep Analysis for synthesis across many papers. Imports accept PDF, DOCX, BibTeX, RIS, PubMed exports and Zotero or Mendeley libraries.
The site names Stanford, Oxford, King's College London, Queen Mary, Brunel and Royal Holloway among institutions using it. Usage is metered in monthly AI credits, from 200 on the free plan up to 100,000.
What it does
- Search over 460 million papers semantically, from a plain-language question
- Turn any research question into an interactive mind map of papers, sub-topics and gaps
- Run a PRISMA-aligned systematic review with an auto-generated flow diagram
- Summarise uploaded papers and chat with them, with answers tied to source paragraphs
- Draft outlines and fully cited sections with Research Agent, from your own library
- Critique a manuscript before submission with Virtual Peer Review
- Manage references end to end — import from Zotero or Mendeley, export in APA, MLA, Chicago or Harvard
When to use ResearchCollab / When not to
A quick filter to help you decide if ResearchCollab is the right fit.
When to use ResearchCollab
- Postgraduate and doctoral students facing their first — or their fortieth — literature review
- Faculty, postdocs and research fellows who need to surface gaps in their field quickly
- Research teams and labs pooling sources across a department with role-based access
- Universities and libraries wanting a vetted AI layer over subscriptions they already pay for
- Analysts running market, policy or legal research across long technical documents
When not to use ResearchCollab
- Developers looking for an API — none is published, and there is no developer documentation
- Anyone who needs a native mobile app, since the product is responsive web only
- Users under 18, who are barred by the terms of use
- Teams that must see a DPA, a subprocessor list or a hosting jurisdiction before signing
- People wanting a general-purpose chatbot for everyday writing rather than sourced synthesis
How to use ResearchCollab
A typical end-to-end flow, from setup to results.
- Try it without an account — the guest Topic Map runs straight from the homepage
- Create a free account at the signup page; no credit card is required
- Type your research question in plain language
- Explore the mind map: sub-topics, related papers, gaps, and paper counts per area
- Create a project and workspace, then add papers by upload, drag-and-drop or batch
- Let the AI extract metadata, summarise, and identify arguments, methods and citations
- Organise the library with folders, sub-folders and custom tags
- Chat with your documents, adjusting summary length or switching on Detailed Summary
- Run the systematic review: fix your protocol and criteria, and let the PRISMA diagram build itself
- Draft with the Outline Generator, critique with Virtual Peer Review, then export fully cited — or hand the whole chain to Research Agent
Pros & Cons
Pros
- A genuinely permanent free plan — 200 credits a month, 40 PDFs, and no card on file
- You can try the Topic Map before creating an account at all
- Every claim links back to a source paragraph, and the vendor explicitly rules out fabricated DOIs
- The whole chain, from question to cited draft, lives in a single product
- PRISMA work is automated rather than tracked by hand in a spreadsheet
- Uploaded documents are stated not to be used for AI training
- Plus and Professional bundle 50+ frontier models and 1:1 human onboarding at a modest price
Cons
- No public API and no developer documentation anywhere on the site
- No native mobile app — responsive web only
- Data governance is thin: no DPA, no subprocessor list, and no hosting country or region disclosed
- No Article 27 EU representative and no DPO named, for a company based outside the EU
- The governing-law clause in the terms is still an unfilled template reading "the laws of [Insert Jurisdiction]"
- Site upkeep is patchy: the About Us page is empty, and three Security sections repeat the same paragraph verbatim
- Free-plan quotas contradict each other between the homepage and the pricing table, and the FAQ still describes "Premium" and "Enterprise" tiers that no longer exist
Pricing & Plans
A permanent free plan is available at no cost and without a credit card. The lowest paid tier, Starter, is priced at USD 9.99 per month on monthly billing; annual billing reduces every paid tier by 20%. All paid plans carry a 30-day money-back guarantee.
- 200 AI credits per month
- 10 AI models
- 200 MB / 40 PDFs
- 1 workspace
- 2 projects
- 50 notes and bookmarks
- limited search and export
- knowledge-base support
- 2
- 500 credits
- 10 Research Agent runs
- Virtual Peer Review
- 3 workspaces
- 5 projects
- 400 PDFs (2 GB)
- 40 AI models
- 25
- 000 credits
- 50 Research Agent runs
- Deep Analysis
- 50+ models including GPT
- Gemini
- Claude
- Perplexity
- 100
- 000 credits
- unlimited Research Agent and unrestricted Deep Analysis
- unlimited workspaces and projects
- 70 GB
- unlimited templates
- advanced Research Trends
- priority computing queues
- 20% cheaper than monthly
Data, GDPR & hosting
A consolidated view of how ResearchCollab handles your data.
GDPR overview
ResearchCollab states GDPR compliance explicitly on its security page, saying data is handled according to the regulation's strict requirements, and claims CCPA compliance alongside it. The privacy policy lists the expected rights — access, update, deletion, restriction, objection and opting out of marketing — but conditions them on your jurisdiction and routes them to privacy@researchcollab.ai. Beyond those claims the evidence thins out. The company is established in the United Kingdom, outside the EU, yet no Article 27 representative is named. No data protection officer is designated, no data processing agreement is published or offered, no subprocessor list exists, and no retention period is stated. The privacy policy even announces a "Last Updated" date that it never displays. Compliance here is asserted rather than documented.
Who owns the data?
Your uploaded papers and notes remain yours. ResearchCollab states that documents are processed only for your own use, are never fed into AI training, and stay workspace-private by default; institutions are told their data is never used for training either. Two caveats sit in the terms. Submitting content grants the company a non-exclusive, worldwide, royalty-free licence to use, reproduce and distribute that content. And the privacy policy allows your information to be transferred as a business asset in a merger or acquisition. Data is not sold or rented, but it may be shared with unnamed third-party service providers, or disclosed where the law requires it.
Reuse rights
You keep full practical use of your own material: papers, notes and Knowledge Bank entries can be exported, and references can be output in APA, MLA, Chicago and Harvard styles. Export is capped on the Free and Starter plans, where only PDF export is offered, and unrestricted on Plus and Professional. AI conversations are stored for your own reference and can be deleted at any time, and the privacy policy grants access, correction and deletion rights depending on your jurisdiction, exercised through privacy@researchcollab.ai. The site's own content is a different matter: the terms forbid reproducing, distributing or creating derivative works from it without prior written permission.
Data retention & training
Hosting summary
ResearchCollab documents how it protects data, but not where it keeps it. Encryption is specified: AES-256 at rest and TLS 1.3 in transit, described as keeping stored records unreadable and shielding transfers from eavesdropping. The FAQ adds "secure cloud storage" with continuous monitoring and backup mechanisms, role-based access controls, and periodic security audits and vulnerability assessments. What is missing is jurisdiction. No hosting country, region or cloud provider is named anywhere on the site, no subprocessor list is published, and no data processing agreement is offered. For a research platform holding unpublished work, that is a material gap, and one worth raising with the vendor before any institutional deployment. The company itself is established at 128 City Road, London, United Kingdom. One observation, not a vendor statement: on 23 August 2026 the domain resolved to 34.160.81.203, an anycast node on Google infrastructure geolocated in the United States. That indicates where the website is served from, not where user documents are stored.
Things to keep in mind
Risks and trade-offs to weigh before adopting ResearchCollab.
- Delegating the whole chain to Research Agent can produce a draft you have not actually read, at the cost of understanding your own field
- Students who never learn to search and screen by hand may lose the very research skills the exercise was meant to build
- AI-generated essays and papers can breach an institution's academic integrity rules, whatever the tool itself permits
- A link to a source paragraph proves the citation exists — not that the interpretation drawn from it is sound
- Submitting content grants a broad non-exclusive, worldwide, royalty-free licence under the terms of use
- Your information may be transferred as a business asset if the company is merged or acquired
- Unpublished work sits with a third party whose hosting jurisdiction, subprocessors and retention periods are all undisclosed
Setup & Integrations
Technical difficulty
Very low. Nothing is installed: ResearchCollab is a web application, and the guest Topic Map runs before you create an account at all. Signing up is free and needs no card, and the first useful result comes from typing a question. Papers are added by drag-and-drop, or imported from Zotero and Mendeley. An optional browser extension saves material from the web. Plus and Professional include a 1:1 onboarding session in which templates are built around your methodology. Institutions need vendor involvement for SSO and IP-based access. The only real prerequisite is machine-readable PDFs: scanned documents must be OCR'd first.
Deployment
Integrations
Behind ResearchCollab
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
Is there a free plan, or only a trial?
How reliable are the citations?
Are my uploaded papers used to train the AI?
What can I import, and how much can I store?
Is there an API?
Is there a mobile app?
What does the systematic review actually automate?
Where is my data hosted, and for how long is it kept?
What are AI credits, and what happens if I run out?
Is there a minimum age?
Should you pick ResearchCollab?
ResearchCollab covers an unusually complete arc for a young product: you arrive with a question and leave with a cited draft, and the steps between — mapping the literature, screening it against a fixed protocol, synthesising and critiquing — all happen in one workspace. The PRISMA automation is the most concrete thing on offer. Generating the flow diagram and tracking inclusion decisions removes a genuinely tedious, error-prone chore, and the Evidence Tracker's habit of tying each claim to a source paragraph is the right answer to the citation problem that makes general chatbots unusable for academic work.
The commercial terms are fair. A permanent free plan with no card, a guest mode needing no account, and a USD 9.99 entry tier bundling frontier models make it easy to test the claims yourself — which matters, because the headline figures (460 million papers, the named universities, the quoted Oxford researcher) are vendor assertions not independently verified here.
The reservations are about maturity rather than capability. Data governance is asserted more than documented: GDPR and CCPA compliance are claimed, but there is no DPA, no subprocessor list, no stated retention period, no hosting jurisdiction, and no Article 27 representative for a UK company selling into Europe. The site shows its age in places too — an empty About Us page, three identical paragraphs on the security page, a governing-law clause still reading [Insert Jurisdiction], and free-plan quotas that disagree between two pages.
For a student or researcher on the free tier, the risk is small and the time saved may be real. For a lab or institution with compliance obligations, the paperwork that would normally accompany a purchase of this kind is not yet there — and that is the first question to put to the vendor.
- Choosing a selection results in a full page refresh.
- Opens in a new window.