arXiv
arXiv
Open repository of research manuscripts and their versions, independent since 2026; moderation is not peer review.
Read the record →Visit arxiv.org ↗Search engines, AI assistants and research tools. Explore 22 selected destinations with their publisher and official domain.
Search engines, AI assistants and research tools. Each listed website retains its own publisher and official destination. Read the individual records in English or switch to their French counterpart.
Compare two or three resources →
Browse each subject, compare the resources and open an English record with publisher details and practical checks.
Rights, official procedures, public data and French institutions.
Explore this subject → 14 websites · bilingual recordsHealth information, prevention, care and support resources.
Explore this subject → 14 websites · bilingual recordsCareers, job search, training, business creation and management.
Explore this subject → 30 websites · bilingual recordsStudy guidance, courses, libraries and open knowledge.
Explore this subject → 17 websites · bilingual recordsGeneral news, public broadcasting and media literacy.
Explore this subject → 14 websites · bilingual recordsMuseums, archives, creative work, events and cultural heritage.
Explore this subject → 22 websites · bilingual recordsSearch engines, AI assistants and research tools.
Explore this subject → 40 websites · bilingual recordsWeb standards, software, development, archives and secure communication.
Explore this subject → 13 websites · bilingual recordsTransport, maps, destinations and journey planning.
Explore this subject → 15 websites · bilingual recordsPurchases, energy, housing, renovation and consumer protection.
Explore this subject → 17 websites · bilingual recordsScientific research, climate, weather, geography and natural risks.
Explore this subject → 12 websites · bilingual recordsInternational institutions, global data and cooperation.
Explore this subject → 10 websites · bilingual recordsSports organisations, competitions and outdoor activities.
Explore this subject →Search engines, AI assistants and research tools. Each record identifies its publisher and official destination, with related resources and practical checks.
arXiv
Open repository of research manuscripts and their versions, independent since 2026; moderation is not peer review.
Read the record →Visit arxiv.org ↗Microsoft
Microsoft's search engine for web pages, images, news and AI assisted features.
Read the record →Visit bing.com ↗Brave Software
Web search engine with its own index, plus image and news search.
Read the record →Visit search.brave.com ↗OpenAI
Conversational assistant for explanations, ideas, analysis and drafting; verify important outputs.
Read the record →Visit chatgpt.com ↗Anthropic
Conversational assistant for questions, writing, code and documents.
Read the record →Visit claude.ai ↗Consensus
Search research papers with assisted summaries; read the studies and methods before relying on an answer.
Read the record →Visit consensus.app ↗DuckDuckGo
Search engine focused on less personalised results and shortcuts to other websites.
Read the record →Visit duckduckgo.com ↗Ecosia
Search engine that reports using part of its revenue for climate projects.
Read the record →Visit ecosia.org ↗Google's AI assistant for discussion, explanation, summaries and work across formats.
Read the record →Visit gemini.google.com ↗General web search with images, news, maps and related services.
Read the record →Visit google.fr ↗Search for academic papers, books, theses, citations and other scholarly work.
Read the record →Visit scholar.google.com ↗Microsoft
Official access to Copilot; check sources and features available for your account.
Read the record →Visit copilot.microsoft.com ↗Mojeek
Search engine with its own index; compare results and review its settings.
Read the record →Visit mojeek.com ↗Perplexity AI
Conversational answer service with source links to open and verify.
Read the record →Visit perplexity.ai ↗Qwant
European search engine; compare results by language, topic and content type.
Read the record →Visit qwant.com ↗SciSpace
Assisted search and reading of research papers, with links to original publications.
Read the record →Visit scispace.com ↗Allen Institute for AI
AI assisted discovery of papers, authors, citations and research topics.
Read the record →Visit semanticscholar.org ↗Startpage
Search engine highlighting personal data protection.
Read the record →Visit startpage.com ↗Mistral AI
Mistral AI’s assistant and work tools, now presented as Vibe, formerly Le Chat.
Read the record →Visit chat.mistral.ai ↗Wolfram Research
Computational knowledge tool for calculations, data exploration and structured queries.
Read the record →Visit wolframalpha.com ↗Yahoo
Yahoo search portal; distinguish search results, ads and other portal content.
Read the record →Visit fr.search.yahoo.com ↗You.com
Web search, content extraction and answer APIs for applications and AI agents.
Read the record →Visit you.com ↗Try a broader name, publisher or subject.
These websites belong to our network. Each has its own editorial scope; check its current information before relying on it.
Crossref relationships have a direction: described and linked objects play different roles. Read the relationship before replacing a reference.
| Situation | What changes | Useful check |
|---|---|---|
| isPreprintOf / hasPreprint | Links a preprint and associated publication in opposite directions. | Check which document was actually read. |
| isVersionOf / hasVersion | Declares a version relationship. | Compare tables and conclusions; the link does not mean identical text. |
| isReviewOf / hasReview | Links a review and its subject. | Read the review and its scope before concluding. |
A working DOI identifies a resource without awarding a quality label. A missing metadata relationship does not establish that works are unrelated. Retain the title, version and passage used.
Documentation : Crossref — Relationships ↗ · · Read the reference record →
Define the expected result before choosing a search engine, calculator or assistant. Useful answers remain connected to documents you opened and explicit assumptions.
A version suffix distinguishes successive states of an arXiv submission.
A question to resolve: Your notes concern v1, but the unversioned link now opens v3. What should you retain?
| Task | Information to retain | Starting point |
|---|---|---|
| Locate publications | Query, title, authors, version | Google Scholar |
| Read a preprint | Version, status, later publication | arXiv |
| Check a calculation | Variables, units, assumptions, rounding | WolframAlpha |
| Check a sourced answer | Citation, passage, date and context | Perplexity |
| Prepare a literature review | Query, filters and selected references | SciSpace |
Keep the question, date, opened sources, relevant passages and unconfirmed points. Remove confidential information before submitting it to an external tool.
A summary seems useful. Reconstruct the chain connecting the claim to a document and its supporting passage.
Record title, authors, date and identifier. Distinguish automatic summaries from author-written text.
Check preprint, published paper or new version. Record relevant changes.
Read method, result and nearby limitations. Keep conditions associated with the conclusion.
A claim accompanied by an identified document, version and checked passage.
A citation in an automatic answer does not establish that it supports the claim.
Links make checking possible. Open them and confirm that they support the claims, period and context.
For verifiable facts, cite the original document you consulted. Use the assistant for preparation and preserve remaining uncertainties.
No. Keep executed queries and decisions, then verify original documents.
These resources answer different questions. Match one to your task, then make the suggested check before relying on its result.
Search by title, author and bibliographic variants.
Check first Open the publisher or repository to check the text’s version.
Find Google Scholar →Follow authors, citations and topics around a publication.
Check first Read the studies before inferring a consensus.
Find Semantic Scholar →Use another search index to find sources absent from an initial list.
Check first Compare domains and distinguish results from ads.
Find Brave Search →Ask for assumptions and a research plan.
Check first Verify important claims against an accessible, dated source.
Find ChatGPT →These comparisons show when two resources in the same subject answer different questions.
arXiv hosts versions of research papers; Consensus helps explore studies around a question. Open the papers before accepting a conclusion and check their publication status.
WolframAlpha can explore calculations and conversions with parameters; a conversational assistant can help frame and explain. In both cases, inspect assumptions and data sources.
The portal gives you context before the external click. Current conditions and sensitive instructions must still be confirmed with the destination publisher.
Start with the category that matches your practical need rather than the name of a platform.
Use the publisher, description, country and topic tags to understand what the resource covers.
Confirm that the external domain matches the publisher and service you intended to reach.
Review current dates, prices, availability, terms and sensitive instructions on the destination website.
The directory contains 232 useful starting-point websites across 13 subjects, including public services, health, education, media, culture, search, digital tools, travel, consumer information, science and sport.
No. Inclusion is an editorial orientation choice, not a technical certification, commercial ranking or guarantee covering every page and service on the external website.
Yes. Every directory record has an English page with a description, publisher, official destination, a specific use case and practical checks. Use the language menu to open its French counterpart.
Check the domain, publisher and purpose described in the record, then confirm current terms, dates and sensitive information on the destination website before acting.
Use the Contact page and provide the relevant URL, the inaccurate passage and a supporting source when possible.