Recall
Recall is a personal AI knowledge base that saves articles, YouTube videos, podcasts, PDFs and notes in one click, then summarises, tags and links them automatically so you can chat with everything you have learned.
What is Recall?
Recall is a self-organising personal knowledge base built around AI. You save an article, a YouTube video, a podcast, a PDF, a tweet, a recipe or a Wikipedia page in a single click, and the tool turns it into what it calls a Recall Card: an automatic summary, extracted keywords, smart tags, and links to everything related you have already stored. Nothing has to be filed or connected by hand, which is precisely the argument the product makes against manual note-taking tools.
Four other layers sit around that library. AI chat answers questions from your own saved material, from the open web, or from both at once, and you choose the model yourself among ChatGPT, Claude, Gemini, Grok and DeepSeek, switching mid-conversation if you want. Augmented Browsing reads the page you are on, spots concepts you have already stored and resurfaces them, with the processing done locally on your own device. Quizzes on a spaced-repetition schedule move what you saved into long-term memory. Listen Mode reads summaries aloud in a voice you can clone.
A full block editor handles your own writing, with tables, to-do lists and code blocks, so Recall works as a note-taking app rather than only a read-later queue. The stated processing limits are generous: YouTube videos up to ten hours, PDFs up to 300 pages, and translation of summaries and quizzes into ten languages.
The company positions the product as one replacement for several tools at once, borrowing Notion's editor, Obsidian's graph, NotebookLM's source-grounded chat and Perplexity's research depth while dropping the manual setup and the source limits. It claims more than 700,000 users and displays a 4.7 out of 5 rating from 464 reviews.
Recall was founded in 2022 and is run from Amsterdam by Paul Richards (CEO), Igor Gligorevic (CTO) and Sankari Nair (COO). The site takes care to distance itself from two similarly named products: Microsoft Recall, the Windows 11 screenshot feature, and Recall.ai, a meeting-recording service. Recall is strictly single-user; the terms forbid sharing an account.
What it does
- Save any article, video, podcast, PDF or web page in a single click
- Generate an instant AI summary, concise or detailed, with timestamps on video
- Chat with your own saved knowledge, the open web, or both at once
- Choose and switch the AI model, from ChatGPT to Claude, Gemini, Grok or DeepSeek
- Resurface related saved material automatically while you browse
- Revise through generated quizzes on a spaced-repetition schedule
- Export the whole knowledge base to Markdown at any time
When to use Recall / When not to
A quick filter to help you decide if Recall is the right fit.
When to use Recall
- Researchers and academics managing literature, papers and field notes across long periods
- Students revising with automatic summaries, quizzes and spaced repetition
- Founders and operators tracking competitive landscapes and industry reading
- Writers, journalists and creators building a searchable reference library
- Lifelong learners who watch, read and listen to far more than they can retain
When not to use Recall
- Teams needing a shared workspace: accounts are individual only and sharing is prohibited
- Anyone looking for screen recording, which is Microsoft Recall, a different product
- Anyone looking for meeting recording and transcription, which is Recall.ai, another product
- Organisations requiring a DPA, a named legal entity or a security certification before buying
- Very high-volume automated users, whom the Fair Use Policy explicitly targets
How to use Recall
A typical end-to-end flow, from setup to results.
- Create a free account at app.recall.it, with no credit card required
- Install the browser extension for Chrome, Firefox, Edge or Safari, the fastest way to save
- Add the iOS or Android app if you want to capture and read on the move
- Save your first item in one click from the browser, the phone or the web app
- Let Recall build the Recall Card: summary, smart tags and links to related content
- Edit or refine the summary in the built-in block editor if needed
- Open the chat, choose your source (your knowledge, the web, or both) and your AI model
- Bulk import existing material, up to 1,000 bookmarks or 10,000 Markdown notes
- Generate quizzes and follow the spaced-repetition schedule to retain what matters
- Connect an external assistant through the read-only MCP connector, or export via Settings
Pros & Cons
Pros
- Permanent free plan with unlimited saves and notes, no credit card required
- API and MCP access included from the free tier, which is rare at this price
- No model lock-in: several providers, switchable in the middle of a conversation
- Genuinely automatic organisation, with no tagging or linking work left to the user
- Local-first architecture with EU backup hosting in Belgium and a clear no-training commitment
- Full Markdown export, plus a stated commitment to open-source the project if it shuts down
- Broad platform coverage: four browsers, iOS, Android and a web app
Cons
- No legal entity name, registration number or postal address published anywhere on the site
- A single generic email address covers support, legal and GDPR requests alike
- No DPA, no Article 27 representative and no security certification such as SOC 2 or ISO
- Unlimited is bounded by a Fair Use Policy allowing Recall to pause or adjust a subscription
- Monthly billing costs 20% more than annual on both paid tiers
- Strictly single-user: no collaboration, no shared workspace, no team plan
- Delaware governing law and mandatory arbitration despite an Amsterdam-based team
Pricing & Plans
Recall offers a permanent free plan that requires no credit card. The cheapest paid entry point is the Plus tier at USD 10.00 per month when billed annually, or USD 12.00 per month when billed monthly. A 30-day refund applies to new subscribers.
- 10 AI summaries per month
- unlimited read-later saves
- unlimited notes with blocks
- tables and to-do lists
- note linking and visual knowledge graph
- API and MCP access
- everything in Free plus unlimited AI summaries
- automatic smart tagging
- AI chat across your knowledge and the web
- Listen Mode with custom voices
- automatic knowledge graph
- Augmented Browsing
- quizzes and spaced repetition
- bulk import of up to 1
- everything in Plus plus one-to-one onboarding from the founding team
- free choice of AI model including ChatGPT
- Claude
- Gemini and other frontier models
- and bulk AI actions on up to 100 items at once
Data, GDPR & hosting
A consolidated view of how Recall handles your data.
GDPR overview
Implementation is concrete but incomplete. The privacy policy, last updated on 1 July 2026, carries a dedicated section titled Your Rights (GDPR, CCPA, and similar laws) listing access, rectification, erasure, portability, objection, no automated decision-making, and the Californian opt-out of data sales. Recall states it does not profile, does not make automated decisions and does not sell data. Rights are exercised by emailing support@recall.it, with a stated 30-day response. Backup and sync data is hosted inside the EU, on Google Cloud in Belgium. What is missing is equally clear: no Data Processing Agreement is published or offered, no Article 27 EU representative is designated, no Data Protection Officer is named, and the site never claims GDPR compliance in so many words.
Who owns the data?
The terms and the privacy policy agree on this point: you keep ownership of everything you upload or create, and Recall states it does not access or use your content for any purpose beyond running the service. There is no advertising use, no profiling and no model training on your material. You can access, export or delete your data at any time, with a full Markdown export from Settings and an irreversible account deletion that removes content both locally and from Recall's servers. If you enable the optional MCP connector, the third-party assistant you choose receives card titles, excerpts, source URLs and tags, and that provider's own policy then governs them.
Reuse rights
You may reuse your own content freely and without asking permission: ownership stays with you, and Recall commits to exporting your saved items and summaries in Markdown so the library moves with you whether you stay or leave. On Recall's side the permitted uses are narrow and stated explicitly. Your data serves only to deliver the features you use; it is not used for advertising, profiling or training any AI model, and it is not sold. Browsing activity is not tracked across sites and no browsing history reaches Recall's servers, because Augmented Browsing runs on your device. What you must not do is share, resell or redistribute your account access, which the terms forbid outright.
Data retention & training
Hosting summary
Recall describes itself as a local-first application: your data sits primarily on your own device, and queries and database operations run locally for speed. For backup and syncing across devices, data is stored on Google Cloud in Belgium, in the europe-west1 region, which keeps that copy inside the European Union. Augmented Browsing is processed entirely on the device, so no browsing history reaches Recall's servers and no cross-site tracking takes place. The privacy policy names the third parties involved in running the service: Google Analytics for usage analytics, Sentry for error tracking, Google Firebase for cloud storage and database services, PostHog for website analytics, and OpenAI when you connect ChatGPT through the MCP connector. No hosting certification such as SOC 2 or ISO 27001 is mentioned anywhere, and no Data Processing Agreement is published or offered.
Things to keep in mind
Risks and trade-offs to weigh before adopting Recall.
- Outsourcing memory to a tool can weaken the very recall it promises: the quizzes exist precisely because saving is not learning
- Instant summaries invite you to skip the source, and a summary of a summary is where nuance and counter-argument quietly disappear
- No legal entity, registration number or postal address is published, so you would have little to point at in a dispute
- Unlimited is contractual, not absolute: the Fair Use Policy lets Recall pause or modify a subscription it judges excessive
- The optional MCP connector sends card titles, excerpts, source URLs and tags to a third-party assistant, governed from then on by that provider's policy
- A single generic email address handles support, legal and GDPR matters, with no dedicated privacy contact and no DPA available
- Building a lifelong library inside one vendor creates real switching costs, mitigated but not removed by the Markdown export
Setup & Integrations
Technical difficulty
Very low. Signing up is free and needs no credit card, and the web app works immediately with nothing to install. The browser extension, available for Chrome, Firefox, Edge and Safari, is the recommended step and takes a moment. Because tagging, linking and the knowledge graph are automatic, there is no taxonomy to design and no configuration to plan. Only the optional API and MCP connector, whose keys live in Settings then Integrations, call for any technical comfort. Max subscribers additionally receive one-to-one onboarding from the founding team.
Deployment
Apps stores
Integrations
Supported languages
Behind Recall
Fundraising
Social
Resources
All the official URLs gathered for verification and reference.
Alternatives
Tools that compete with or complement Recall.
Frequently asked questions
Does Recall train AI models on my content?
Where is my data stored?
Is the free plan really free?
Which AI models can I use?
What content can Recall handle?
Which languages does Recall support?
Can I export or delete my data?
Is there a refund or a student discount?
What happens if Recall shuts down?
Is this Microsoft Recall or Recall.ai?
Should you pick Recall?
Recall does one thing with unusual thoroughness: it removes the filing work from personal knowledge management. Summaries, tags, connections and the knowledge graph all happen without you touching them, and the payoff is real for anyone who consumes far more long-form content than they can remember. The multi-model chat is a genuine differentiator, since you are not tied to whichever provider the vendor happened to sign with, and you can change model in the middle of a conversation.
The data posture is better than most tools of this size. Recall states plainly that it never trains models on customer content, keeps Augmented Browsing entirely on the device, backs up to Google Cloud in Belgium rather than outside the EU, and lets you export everything to Markdown at any moment. The team even commits to open-sourcing the project if the company ever shuts down.
The reservations are about the company, not the product. No legal entity name, no registration number and no postal address appear anywhere on the site, and a single generic address handles support, legal and GDPR requests alike. There is no DPA, no Article 27 representative and no security certification on offer, which will stop some organisations from adopting it. The terms also send disputes to Delaware arbitration even though the team sits in Amsterdam.
Two practical caveats: unlimited is bounded by a Fair Use Policy that allows Recall to pause or adjust a subscription, and the monthly price is 20% above the annual one on both paid tiers. Recall suits individual researchers, students, founders and writers working alone. Teams that need shared workspaces, and buyers who need a named counterparty on paper, should look elsewhere.
- Choosing a selection results in a full page refresh.
- Opens in a new window.