# Notebooker documentation Save what you find, ask questions about it, and build something out of it. Notebooker answers from your own sources, reads them aloud as podcasts, and turns them into study material. Source: https://notebooker.ai/docs/ ================================================================================ # Apps, devices & syncing Source: https://notebooker.ai/docs/apps/ Notebooker is one library you can reach from anywhere. Save on your phone, ask a question on your laptop, listen on a walk: it is all the same notebook, kept in sync automatically. Where Notebooker runs Web. app.notebooker.ai in any browser, no install required. iOS and Android. Native apps with quick save and podcast generation, from the App Store and Google Play. Native desktop. macOS (Apple Silicon and Intel), Windows, and Linux builds. Browser extensions. One-click save for Chrome, Edge, Brave, and Firefox, plus a Safari extension in beta. No install? A bookmarklet works in every browser. Grab any of them from the download page. On macOS you can also install the desktop app with Homebrew: brew install --cask notebooker-ai/tap/notebooker How syncing works Everything you save, generate, or grade is stored against your account, not on a single device. Save a link from the iOS share sheet and it is waiting in the web app seconds later. Flashcard reviews, podcast subscriptions, and notebook changes all follow you from device to device. If you have connected your own storage, your files land in your bucket while the sync metadata keeps every device pointed at the same library. Related Download Notebooker Importing & bulk actions Bring your own storage ================================================================================ # Bring your own storage Source: https://notebooker.ai/docs/storage/ By default, the files Notebooker creates for you - uploaded sources, podcast audio, and creator artifacts - live in Notebooker-managed storage, organized privately under your account. If you’d rather keep your own copy of everything, you can connect your own object storage instead. Why bring your own storage? You own the data. Files land directly in a bucket you control. Portability. Keep your podcasts and documents in infrastructure you already use. Less metered storage. Notebooker-managed storage is free for your first 5 GB; only bytes above that are billed, at $0.10 per GB-month, prorated daily, against your credit. Files in your own bucket aren’t Notebooker storage, so they’re never metered. No surprises at deletion. When you delete your Notebooker account, your own bucket is never touched - Notebooker only cleans up its own storage. Supported providers Any S3-compatible object storage works, including: Amazon S3 Cloudflare R2 Backblaze B2 DigitalOcean Spaces MinIO (self-hosted) Other S3-compatible services (via a custom endpoint) Set it up Open Settings → Storage (S3) in Notebooker and fill in: Field Required Notes Access key ID required From your storage provider. Secret access key required Stored encrypted; never shown again. Bucket name required The bucket new files are written to. Region optional Defaults to us-east-1. Endpoint URL optional Needed for R2, B2, Spaces, MinIO, etc. (anything that isn’t AWS S3). Public URL optional A CDN/base URL if you serve files from a custom domain. Path-style addressing optional Turn on for MinIO and some S3-compatible services. Then: Click Test connection to confirm the credentials and bucket work. Click Save. Your credentials are encrypted at rest, and each request only ever uses your own credentials for your own files. Note: Bringing your own storage applies to new files written after you connect it. Files already stored in Notebooker’s bucket stay there until you export or delete them (below). If you wipe Notebooker’s storage before migrating, records that still point at those files will break - so connect your bucket first. Download your data You can export everything Notebooker has stored for you at any time: Go to Settings. Under storage, choose Export my files. Notebooker zips every file stored on your behalf and emails you a download link. The link is valid for 7 days. Large libraries can take a few minutes to package - you’ll get the email when it’s ready. Delete your data Two separate actions, depending on what you want to remove: Delete your Notebooker files (keep your account) Under Settings → storage, Delete my Notebooker files permanently removes everything stored in Notebooker’s bucket for your account. You’ll be asked to type a confirmation phrase. If you haven’t migrated to your own storage, the notebook entries pointing at those files will no longer be able to load them. Export first if you want a copy. Delete your account Settings → Delete my account removes your account entirely. This: Cancels any active subscription. Deletes your account record, sessions, and billing history. Purges your files from Notebooker’s storage and your sources/episodes from the backend. Leaves your own bring-your-own-storage bucket untouched - Notebooker never deletes files from a bucket you own. Account deletion is permanent and can’t be undone, so export anything you want to keep first. Related Subscribe to your podcast feed Privacy Policy ================================================================================ # Choose your AI models Source: https://notebooker.ai/docs/models/ Notebooker comes with its own first-party AI models set as the defaults, so everything works the moment you sign up. When you want more control over what reads your material, you can point Notebooker at a different provider, bring your own keys, or run a local model instead. Your options Notebooker’s built-in models (default). Nothing to configure. Your plan’s AI credit covers them. Your own provider keys. Add an API key for OpenAI, Anthropic, or another supported provider and Notebooker uses your account directly, billed by that provider rather than through Notebooker. Only Notebooker AI and service usage counts against your credit - requests on your own keys never do. A local model. Point Notebooker at a model running on your own machine or network so your material never leaves it. None of ours. Choose providers so that Notebooker’s own models never see your data. “Your keys, your models” is a setting, not a promise. Set it up Open Settings and find the AI or Models section. For a hosted provider, paste the API key and pick which model to use for chat, insights, and creators. Keys are stored encrypted and only ever used for your own requests. You can mix providers: one model for chat, another for podcasts, a third for embeddings. If a provider is down or a key is missing, Notebooker falls back to a working default so you are never stuck. Run your own endpoint Want a model that isn’t on the built-in list, or your own OpenAI-compatible gateway? See Run your own AI provider to deploy an endpoint with cloudflare-ai-deployer and point Notebooker at it. Related Run your own AI provider How Notebooker handles content Bring your own storage ================================================================================ # Citations & answer coverage Source: https://notebooker.ai/docs/citations/ Ask Notebooker a question and the answer isn’t just prose - every claim carries a citation that links back to the exact source, note, or insight it came from. Coverage indicators go a step further and show you how much of the retrieved material actually made it into the answer, so you can judge an answer instead of trusting it blindly. Citations Inline citations render as numbered references ([1], [2]) with a reference list under the answer. Click any of them to open the cited source, note, or insight and read the original in full. Citations appear in: Chat - grounded in the sources you selected into the conversation’s context. Ask - grounded in whatever the automatic searches retrieved across your whole library. Wiki creations - every topic article the Wiki creator writes cites the sources it drew from, as footnotes. Answer coverage The common failure of research assistants is silent: an answer sounds complete, but the model only used two of your twelve documents. Notebooker surfaces this instead of hiding it. In Ask - the final answer ends with a coverage strip: “Cited 4 of 7 documents consulted” - how many retrieved documents the answer actually cites. A clickable “Consulted, not cited” list - documents the searches retrieved that didn’t make it into the answer. If something you expected is in this list, the model saw it and chose not to use it. If it’s absent entirely, the searches never retrieved it - rephrase the question or ask about that document directly. In Chat - each answer shows how many of the items in your selected context it cites. A low count on a broad question is a signal to narrow the question or trim the context selection. Getting better-grounded answers Small context beats big context. Chat sends your selected content in full; a focused selection gives the model less room to wander. Ask for citations explicitly when writing custom prompts - e.g. “cite the specific source for each claim.” Check the not-cited list before acting on an answer that’s supposed to synthesize everything. Related How Notebooker handles content - what gets ingested, embedded, and archived. ================================================================================ # Connect Claude with MCP Source: https://notebooker.ai/docs/mcp/ Notebooker ships a remote MCP (Model Context Protocol) server, so AI assistants that support MCP connectors - Claude on the web, desktop, and mobile among them - can work with your library directly: list your notebooks, add sources, write notes, search everything, and ask questions across your research. There’s nothing to install and no API key to copy. The connector uses OAuth: you sign in to Notebooker once, approve access, and the assistant gets its own revocable credentials for your account. Connecting from Claude In Claude, open Settings → Connectors and choose Add custom connector. Enter the connector URL: https://app.notebooker.ai/mcp Claude opens a Notebooker sign-in window. Log in (if you aren’t already) and click Authorize on the consent screen. That’s it - Claude can now see the Notebooker tools. Try “list my notebooks” or “add this article to my Reading notebook”. Any other MCP client that supports remote servers with OAuth works the same way - point it at the same URL and complete the sign-in flow. To disconnect, remove the connector in your client’s settings. The client’s access tokens expire on their own within a day and refresh only while the connector stays installed. What the connector can do Tool What it does manage-notebooks List, create, rename, archive, or delete notebooks. manage-sources List sources, add links or pasted text, check processing status, retitle, retry, or delete. manage-notes List, read, write, update, or delete notes. search Keyword or semantic search across your sources and notes. ask Ask a question across your whole library and get a cited AI answer. chat Chat with a specific notebook’s sources, with retrieval. transformations List your AI transformations or run one on text. insights Generate, read, and save AI insights on sources. A few things worth knowing: Adding sources is asynchronous. When the assistant adds a link, Notebooker fetches, extracts, and embeds it in the background - the same pipeline as adding a source in the app. The assistant gets a status tool to poll until processing finishes. Everything is scoped to your account. The connector sees exactly the notebooks, sources, and notes your login sees - nothing else. AI actions draw on your credit. Asking questions, chatting, running transformations, and semantic search use your monthly AI credit exactly like the same actions in the app - see Usage, credit & budgets. Reading and searching by keyword are free, and reads keep working even when your credit runs out. Deletes are real. If you ask the assistant to delete a notebook or source, it’s deleted - same as doing it in the app. Claude asks you to confirm tool calls that change data, so read those prompts before approving. MCP or the API? Use the MCP connector when an AI assistant should work with your library conversationally. If you want to drive Notebooker from scripts, n8n, or anything that speaks plain HTTP, personal API keys are the better fit. ================================================================================ # Creator plugins Source: https://notebooker.ai/docs/creators/ Creators turn the sources in a notebook into a finished artifact you can study, present, or share. Pick a creator, set a few options, and Notebooker generates it from your content. Each creator is powered by a well-known open-source library, and several let you choose exactly what kind of output you want. At a glance Creator Produces You can download Powered by Flashcards Q&A study cards Anki .apkg genanki Charts Data charts - (interactive) Flint (Vega-Lite / ECharts / Chart.js) Infographics Diagrams & layouts - (interactive) AntV Infographic Mind maps Hierarchical map Markdown / PNG / SVG Mermaid mindmap syntax Textbooks Multi-chapter book HTML / PDF / EPUB Quarto Essays Structured essay HTML / PDF Quarto Study guides Topic-by-topic guide HTML / PDF Quarto Slideshows Slide deck HTML / PPTX / PDF reveal.js + Quarto Timelines Interactive timeline - (interactive) Custom renderer Wikis Interlinked topic articles Obsidian-ready Markdown + zip Custom renderer Socratic tutor Guided tutoring session Printable study sheet (Markdown) Custom renderer Counterpoints Two-sided debates of your material Markdown / PDF Custom renderer + Quarto Websites Multi-page static website Site zip - or publish it live Quarto websites Flashcards Generates question-and-answer cards from your notebook, with a built-in study mode that does real spaced repetition: grade each card Again / Hard / Good / Easy and the FSRS scheduler decides when it comes back, with your progress saved to your account. Export the deck as an Anki .apkg to study in Anki or any compatible app - see Flashcards & spaced repetition. Options: number of cards (1–100, default 15), deck name (pre-filled from the notebook’s name). Library: genanki for the Anki export; FSRS for scheduling. Charts Generates data charts: the model designs each chart as a declarative spec, rendered by your choice of three charting libraries - Vega-Lite, ECharts, or Chart.js. Pick a library and chart type in the creation dialog, or leave either on Auto and let Notebooker choose the best fit for the data. Chart types: each library has its own catalog - the classics (line, bar, area, scatter, pie) plus library-specific types like heatmaps, radar charts, funnels, treemaps, sunbursts, and gauges. Options: rendering library (auto / Vega-Lite / ECharts / Chart.js), chart type (auto or a specific type), theme (auto, light, dark), how many charts (1–6). Library: Flint - browse the gallery to see every chart type in each library. Infographics Diagrams that explain relationships and structure rather than plot numbers, rendered with the AntV Infographic engine. Graphic families: Lists - bullets, grids, checklists, waterfalls, zig-zags. Sequences - steps, stairs, funnels, pyramids, roadmaps, timelines. Comparisons - side-by-side, quadrants, SWOT. Hierarchies - org charts, trees, mind-map-style branches. Relations - node-and-edge flow diagrams. Options: theme (auto, light, dark, hand-drawn), how many (1–6). Library: AntV Infographic - examples gallery. Mind maps Generates a single-root mind map of your material. Export to Markdown, PNG, or SVG. Options: maximum depth (2–8, default 5), how many (1–6). Format: Mermaid mindmap syntax under the hood. Textbooks Generates a full multi-chapter textbook from your notebook, optionally with practice exercises. Options: number of chapters (1–20), audience (high school / undergraduate / graduate / general), include exercises, output formats. Download: HTML, PDF, and/or EPUB. Library: Quarto (with the Tectonic LaTeX engine for PDF). Essays Generates a structured essay in a style of your choosing. Options: variant (argumentative / expository / comparative), length (short / medium / long), output formats. Download: HTML and/or PDF. Library: Quarto. Study guides Generates a study guide that walks topic by topic: key concepts, what you need to know, common misconceptions, and a glossary. Options: number of topics (1–20), audience, output formats. Download: HTML and/or PDF. Library: Quarto. Slideshows Generates a slide deck - including data slides whose charts and infographics are pre-rendered with the same AntV engine as the Infographics creator. Slide types: title, section, bullets, image+text, chart, infographic. Options: number of slides (3–40), theme (reveal.js themes incl. light, dark, serif, moon, solarized, night, league), output formats. Download: HTML (interactive reveal.js), PPTX (PowerPoint), and/or PDF. Libraries: reveal.js + Quarto, with AntV Infographic for data slides. Timelines Generates an interactive timeline of events, optionally grouped into labeled lanes. Handles plain years, full dates, and date-times. Item types: point, range, box, background. Options: maximum events (3–60), grouped lanes on/off. Wikis Distills your notebook into a set of interlinked, Wikipedia-style topic articles: Notebooker picks the topics, writes a cited article for each (every claim footnoted back to the source it came from), and cross-links related articles with [[wikilinks]]. Browse the wiki right in the app - topic index, navigable links, references - or download it as Obsidian-ready Markdown: one .md per topic with YAML frontmatter, footnote citations, an index note, and a zip you can drop straight into an Obsidian vault. Options: maximum topics (1–20, default 8), article length (concise / standard / comprehensive), index note on/off. Download: per-topic Markdown files and the vault zip. Socratic tutor Turns your notebook into a guided tutoring session that teaches by asking, not telling. Notebooker plans the key concepts in your sources, then writes a full dialogue for each one: an opening question, a checklist to grade your own answer against, “I think…” buttons for the tempting wrong turns (each answered with a counter-question, never a correction), three hints that break the problem into smaller questions, and - only when you’ve used them all - a model answer cited back to your sources. Work through it in the app; every question is answerable from your material alone. See Socratic tutoring for how a session flows. Options: number of concepts (3–25, default 10), difficulty (recall / application / synthesis - synthesis adds capstone questions that combine concepts), tutor persona, show model answers on/off, scope (e.g. “only through chapter 5” to keep later material out). Download: a printable study sheet - questions and hints up front, answers at the back. Counterpoints Finds the genuinely debatable questions in your material and argues both sides properly. For each issue you get a neutrally-phrased question, two sides labeled from the content itself (“Growth first” vs “Stability first” rather than a generic For/Against), and matched argument pairs: a point with its evidence, a counterpoint that rebuts that specific point, and a steel-manned response - concessions allowed, strawmen not. An optional synthesis names the crux of each disagreement and what evidence would settle it, without declaring a winner. Read the debates in the app’s two-column view, or download them. Options: number of issues (1–10, default 3), points per issue (2–8, default 4), synthesis on/off, output formats. Download: Markdown, and/or PDF. Websites Turns your notebook into a complete multi-page static website: a home page that orients a first-time visitor plus a page per topic, cross-linked, with a navigation bar, a collapsible sidebar, and built-in search. Every page is planned from your sources and written only from your material. Download the site as a zip and drop it on any static host (GitHub Pages, Netlify, Cloudflare) - or, if you’ve connected your own storage with a public URL, publish it live in one click and get a shareable link. See Websites & publishing for details. Options: number of pages (2–15, default 6), theme (10 Bootswatch looks, from clean cosmo to dark darkly), search on/off, sidebar on/off. Download: the full site as website.zip - and a live URL when published. Library: Quarto websites - the same engine behind quarto.org itself and course.fast.ai. Tips Steer the output in your instructions. For infographics especially, naming the layout you want (“lay this out as a SWOT”) gets you the graphic you have in mind - browse the AntV examples gallery for ideas. For charts, pick the library and chart type right in the creation dialog, or browse the Flint gallery. Pick formats you’ll actually use. Document creators (textbook, essay, study guide, slideshow) let you choose which files to produce - fewer formats generate faster. Bigger context, richer creations. Notebooker’s built-in models are chosen to be economical, so they work within smaller context windows. When a notebook is larger than the model’s limit, Notebooker automatically distills it to fit - condensing and pruning your sources before generating. Nothing breaks, but the creator is working from a summary rather than the full detail. If you want a creator to really draw on everything you’ve saved, bring your own AI keys and pick a model with a larger context limit - requests on your own keys are billed by your provider, not your Notebooker credit. The more of your notebook the model can see at once, the more thorough and specific the result. ================================================================================ # Export your notes & decks Source: https://notebooker.ai/docs/exporting/ Everything Notebooker makes for you is yours to take elsewhere. Three exports cover the common destinations: A single note → Markdown Any note can be downloaded as a .md file. The filename carries the note’s title and the body is the note exactly as written - no wrapper, no metadata to strip. A whole notebook → zip of Markdown Export a notebook and you get a zip with one Markdown file per note, named from the note titles. Drop the folder straight into an Obsidian vault, Logseq, or any Markdown editor and everything just opens. API routes, if you’re scripting it with an API key: GET https://app.notebooker.ai/api/v1/notebooks/{notebook_id}/export/markdown → notebook-notes.zip GET https://app.notebooker.ai/api/v1/notes/{note_id}/download → note.md Full details in the API reference. Flashcards → Anki Every flashcards artifact includes an Anki .apkg download. Identifiers are stable, so re-importing an updated deck updates cards in place instead of duplicating them. See Flashcards & spaced repetition. Everything else Textbooks, essays, study guides, slideshows download as HTML / PDF (and EPUB or PPTX where it makes sense) - see Creator plugins. Podcast episodes are plain MP3s in your private podcast feed. Your entire storage - every file Notebooker holds for you - exports as one zip from Settings, and you can bring your own bucket so files never live anywhere else to begin with. ================================================================================ # Flashcards & spaced repetition Source: https://notebooker.ai/docs/flashcards/ The Flashcards creator turns a notebook into a deck of question-and-answer cards. The built-in study mode is a real spaced-repetition system, not just shuffle-and-reveal: you grade every card, and a scheduler decides when each one comes back. Studying a deck Open a flashcards artifact and the deck view shows what’s waiting: due - cards whose review date has arrived. new - cards you haven’t studied yet. Each card in the list carries a chip: new, due, or in 3d. Hit Study and Notebooker queues due cards first (oldest due leading), then new cards. Reveal the answer, then grade yourself: Grade What it means What happens Again Didn’t remember The card comes back in ~10 minutes, later in this same session. Hard Barely remembered A short interval. Good Remembered A normal interval. Easy Knew it cold A long interval. Each button shows a preview of the next interval (10m, 3d, 2.1mo) before you press it, so you always know what a grade costs. The scheduler Grading runs through FSRS (Free Spaced Repetition Scheduler), the modern algorithm also available in Anki. It tracks a per-card memory model - stability and difficulty - and targets about 90% recall: intervals stretch as a card gets easier for you and shrink when you lapse. Your review state is saved to your account per card, so progress follows you across devices and sessions. If nothing is due, Practice all lets you run the whole deck - early reviews just reschedule cards like any other review. Export to Anki Every flashcards artifact also ships an Anki .apkg deck you can download and import into Anki on desktop or mobile. Deck and card identifiers are stable, so re-importing a regenerated deck updates your existing cards instead of duplicating them. Related Creator plugins - everything else a notebook can become. Export your notes & decks - Markdown, Anki, and full storage export. ================================================================================ # How Notebooker handles content Source: https://notebooker.ai/docs/content-handling/ Notebooker is an educational and research tool. It helps you collect material you have the right to access, organize it into notebooks, and study or reason over it with AI. This guide explains what happens to that content - and, just as important, what never happens to it. You decide what Notebooker saves Everything in your library is there because you put it there - you saved a link, uploaded a file, or imported your own bookmarks. Notebooker works from the specific sources you give it, on your behalf, the way your own browser would when you open a page. It doesn’t roam the web collecting pages you didn’t ask for. Because you choose the sources, you’re responsible for having the right to access and save them. Use Notebooker only for content you’re permitted to access, and in line with the rightful owner’s terms and applicable law - see our Terms of Service. What happens when you add a source When you add a URL or file, Notebooker does a few things, all tied to your account: Extracts the text so you can search, chat with, and summarize it. Renders a PDF snapshot so your notebook keeps a stable copy even if the original page later changes (see Importing & bulk actions). Creates embeddings - numeric representations that power search and chat across your notebook. Retrieve, extract, snapshot, index - so the material is useful to you inside your notebook, and nothing more. Notebooker does not train AI models This is the important part: Notebooker does not train, fine-tune, or otherwise build AI models on your content or on the content you ingest. Your material is used only to deliver the features you ask for - answering your questions, generating a summary or podcast, building a study aid - and nothing else. The default Notebooker AI runs on Cloudflare Workers AI, under terms that don’t use your inputs to train models. If you bring your own AI provider keys, the content you send is handled under that provider’s terms - check them if training or retention matters to you. We don’t sell your content, and we don’t mine it to build models for ourselves or anyone else. Your data stays yours Connect your own S3-compatible storage so archived files live in a bucket you control - see Bring your own storage. Download or delete everything Notebooker stores for you at any time. Deleting your account removes your content and your usage records. Use it the way it’s meant to be used Notebooker is built for learning and research - reading, organizing, studying, and understanding material you have legitimate access to. It isn’t a tool for redistributing other people’s work or for getting around the access conditions a content owner has set. Staying within that keeps you on the right side of both our Terms and the people whose work you’re learning from. ================================================================================ # Importing & bulk actions Source: https://notebooker.ai/docs/importing/ Adding one link to Notebooker is quick. Importing hundreds of them - or running a bulk action across a big selection - is a different story: Notebooker does real work on each source (extracting text, rendering PDF snapshots), and that work runs in the background over time rather than all at once. This guide explains what’s happening, why it can be slow, and how to watch it progress. Import from NotebookLM (Gemini Notebook) Moving from Google’s NotebookLM? The Notebooker browser extension imports a notebook in about a minute - Google offers no export, so the extension does the reading for you: Install the extension (Chrome, Firefox, or Safari) and sign in. Open your notebook on notebooklm.google.com with the Sources panel visible, then click the Notebooker extension icon. The popup shows the notebook’s name and its source list. Adjust the name, untick anything you don’t want, and hit Import. Notebooker recreates the notebook and saves every link-based source through the normal save flow - so each one gets the full treatment: text extraction, indexing for cited answers, and any auto-processing you have enabled. Two honest caveats: Uploaded files and Drive documents can’t be pulled out - Google doesn’t expose them. The importer lists them by name after the import so you can add those files to the new notebook yourself. The importer reads NotebookLM’s page directly, and Google changes that page from time to time. If a scan comes up empty, open the Sources panel, reload the NotebookLM tab, and try again - and if it still can’t read the list, an extension update is usually already on the way. What happens when you import The Import button on your Library lets you upload a bookmarks export - Raindrop or browser HTML, or a Raindrop CSV, up to 50 MB. Tags become notebooks; untagged links are saved on their own. (Bulk import is a premium feature; you can always add single links for free.) When you submit, you’ll see “Import started - we’re adding your bookmarks in the background.” Notebooker then: Turns each distinct tag into a notebook (reusing notebooks of the same name). Splits your bookmarks into small batches and queues them. Creates each source the same way a hand-saved link is created - and kicks off whatever per-source processing you have enabled (see below). Your sources appear in the Library progressively as the batches finish, and you get a single “Bookmark import finished” notification at the end. Heads up on AI credit. Each imported link uses AI credit. If your credit runs out partway through, the remaining links may fail to process - you can re-import them later. For a large import, consider raising your monthly spending cap first so processing isn’t cut off. Bulk actions Select one or more sources in your Library (or use Select all) and a bulk action bar appears. You can: Generate PDF snapshot - queue a PDF render of each selected page. Delete - remove the selected sources. Like importing, these queue work in the background - you’ll get a toast such as “Snapshotting N source(s) - we’ll render and store PDFs shortly” and the items finish over time. The work that runs per source Each source can trigger several background tasks. One of them is substantial: Task What it does Why it takes time PDF snapshot Captures the live webpage as a stored PDF you can open anytime. It opens a real browser and renders the actual page. Snapshots run only a couple at a time, so they drain slowly - especially after a big import. (Plus the usual behind-the-scenes text extraction and indexing for search.) So if you import 100 bookmarks with PDF snapshots turned on, that’s not 100 quick saves - it’s a queued browser render for every link. That fans out across Notebooker’s processing queues and completes over minutes to hours, not seconds. This is deliberate. Imports and bulk actions run on low-priority lanes, so a big batch never slows down a link you save by hand - those stay fast. Watching progress There’s no single progress bar, but each source row tells you where it is: “Processing…” - the source is still being set up. A pulsing document icon - PDF snapshot in progress. When it’s done, the icon becomes View PDF snapshot. You don’t have to keep the page open - processing continues on Notebooker’s servers. Check back and you’ll see more rows finished each time. Make new sources process automatically (or not) Under Settings → Sources you control what happens to newly added sources: Auto-snapshot new link sources - every new URL you add is queued for a PDF snapshot. It’s premium, and it affects only new sources - turning it on doesn’t retro-process your existing library. If you want a large import to finish faster, turn it off before importing so the links come in without the extra snapshot work - then snapshot later, in batches, using the bulk actions whenever you’re ready. Tips for large imports Raise your spending cap first so a long import isn’t cut off when credit runs low. Decide on snapshots up front. Off = fast import of just the links; on = richer results that take longer. Let it run. Close the tab if you like - work continues in the background, and the row status icons show progress when you return. Need account, billing, or troubleshooting help? Visit Support. ================================================================================ # Insight personas Source: https://notebooker.ai/docs/personas/ Transformations run a prompt against a source and save the result as an insight - a summary, an analysis, a reading of the material from a particular angle. Notebooker ships a set of personas so you don’t have to prompt-engineer the classics yourself, and you can add your own. Run them from any source’s Insights panel. The personas Professor - teaches the content instead of summarizing it. Produces lecture notes: learning goals, concepts ordered from fundamentals up with one analogy each, the misconceptions you’re most likely to form, and exam-style questions (without answers) to check yourself. Critic - a frank peer review, deliberately not polite. Names the weak and unsupported claims, what’s conspicuously missing, the strongest good-faith counterargument to the main thesis - and, for calibration, what genuinely holds up. Debate - a debate brief on the source’s central thesis. Steelmans 3–4 arguments for and 3–4 against, writes each side’s best rebuttal, and names the crux the disagreement actually turns on. Beginner’s Guide - re-explains the content for a smart reader with zero background: why it matters, the big picture with one everyday analogy, every piece of jargon defined, then the full story step by step. The workhorses Alongside the personas, the classic transformations are still there: Dense Summary - a rich, compressed summary (applied to new sources by default). Key Insights - the most surprising and useful ideas as tight bullets. Analyze Paper - purpose, contribution, findings, implications, and limitations of a technical paper. Reflections - questions the source should make you ask. Table of Contents / Simple Summary - quick orientation. Make your own Every transformation is just a named prompt - duplicate one, edit the prompt, and it appears in your list next to the built-ins. If a persona is almost right (“Critic, but focused on statistical methods”), copying and tweaking it takes a minute. Related Creator plugins - for artifacts (decks, textbooks, slideshows) rather than insights. Citations & answer coverage - how answers stay grounded in what you saved. ================================================================================ # Knowledge Base Source: https://notebooker.ai/docs/ Welcome to the Notebooker knowledge base. These guides walk through the features that have a few moving parts - how to configure them, what your options are, and where to find more detail. Podcasts & audio Podcast, speaker & episode profiles - control who hosts your podcast, how they sound, and how each episode is structured - including the built-in formats: deep dive, brief, critique, debate, and unabridged. Notebooker TTS voices - listen to every built-in voice, and find the voice catalogs for OpenAI, ElevenLabs, Google, and other providers. Subscribe to your podcast feed - get your private RSS URL and add it to Apple Podcasts, Pocket Casts, Overcast, and more. Study & understanding Search, Ask & Chat - full-text search across your whole library, cited answers over everything with Ask, and close reading over a context you choose with Chat. Flashcards & spaced repetition - grade your cards, let the FSRS scheduler decide when you see them again, and export to Anki. Insight personas - Professor, Critic, Debate, and Beginner’s Guide read your sources the way you need, plus your own custom transformations. Citations & answer coverage - every answer cites your sources, and a coverage meter shows which documents actually informed it. Content & creators Creator plugins - the ten creators that turn a notebook into flashcards, charts, infographics, mind maps, textbooks, essays, study guides, slideshows, timelines, and wikis - including which chart and graphic types you can ask for. Importing & bulk actions - what happens when you import bookmarks or run a bulk action, why it can take a while as Notebooker snapshots pages and extracts text, and how to watch it progress. Watch RSS & YouTube feeds - point a scheduled workflow at a feed or channel and new entries become sources automatically. Your data & automation Apps, devices & syncing - where Notebooker runs (web, iOS, Android, desktop, and browser extensions) and how one library stays in sync across all of them. How Notebooker handles content - what happens when Notebooker ingests and archives the sources you add, why it never trains AI models on your material, and how it’s meant to be used as an educational and research tool. Export your notes & decks - download notes as Markdown, whole notebooks as Obsidian-ready zips, and flashcard decks as Anki files. Usage, credit & budgets - what your monthly credit covers, how pay-as-you-go overage works, and the budget cap that defaults to $0. Bring your own storage - connect your own S3-compatible bucket, and download or delete everything Notebooker stores for you. Webhooks & workflows - trigger Notebooker from other apps (incoming webhooks) and notify other apps when things happen (outgoing webhooks). Personal API keys - drive Notebooker from scripts, n8n, or anything that can send an HTTP request. Connect Claude with MCP - add Notebooker as a custom connector in Claude (or any MCP client) and read, search, and write your notebooks from a chat. Power users Choose your AI models - run on Notebooker’s built-in models, bring your own OpenAI or Anthropic keys, or point at a local model. Run your own AI provider - deploy an OpenAI-compatible endpoint on Cloudflare with cloudflare-ai-deployer and point Notebooker at it. Open source & the Open Notebook platform - the open foundation Notebooker is built on, and where to find the code. Looking for account, billing, or troubleshooting help? Visit Support. ================================================================================ # Notebooker TTS voices Source: https://notebooker.ai/docs/voices/ Notebooker includes its own text-to-speech (TTS) voices, used for podcasts and read-aloud. You can also connect another provider (OpenAI, ElevenLabs, Google, and more) and use their voices instead. Built-in Notebooker voices Notebooker ships 11 voices. melody is the base voice; the other ten are distinct character voices. Assign any of them to a host in a speaker profile. Each sample below is the same sentence so you can compare them directly. Base voice melody - the natural base voice (fastest) Lower / masculine-leaning voices craig nick seth tom tony Higher / feminine-leaning voices belinda deb elisabeth kate kirsten How the voices work melody is generated directly by the speech model, so it’s the fastest and cheapest. The other ten voices start from melody and are run through a voice conversion step to give each its own character - so they take a little longer to generate. All of them are metered the same way as the rest of your Notebooker AI usage. Using another provider’s voices If you’d rather use a commercial TTS provider, add it under Settings → API keys / Models in Notebooker and pick one of its voices. Speech generated on your own key is billed by that provider, not your Notebooker credit - only Notebooker AI and service usage is counted. Notebooker supports these TTS providers - use the linked catalog to find exact voice names: Provider Example voices Full voice catalog OpenAI alloy, echo, fable, onyx, nova, shimmer OpenAI TTS voices ElevenLabs Your account’s voice library (custom + stock) ElevenLabs Voice Library Google Cloud Kore, plus hundreds across languages Google Cloud TTS voices Azure Neural voices per locale Azure speech voices Deepgram aura-2-thalia-en, other Aura voices Deepgram Aura voices Mistral Provider voice set Mistral docs xAI eve, others xAI docs Related Podcast, speaker & episode profiles - assign these voices to your hosts. Run your own AI provider - host the speech model yourself. ================================================================================ # Open source & the Open Notebook platform Source: https://notebooker.ai/docs/open-source/ Most “private by design” tools ask you to trust a policy page. Notebooker is built on top of the open-source Open Notebook platform, so the engine behind your notebooks is something you can actually read. What that means for you You can inspect it. The platform that ingests, embeds, and reasons over your sources is open. Nothing about how your material is handled is a black box. You can fork it. Build on it, adapt it, or run your own copy. You are not locked in. Combined with your own storage and your own AI keys, the important pieces of Notebooker are things you control rather than things you rent. Where the code lives Notebooker on GitHub: github.com/Notebooker-ai The Open Notebook platform: open-notebook.ai Bring-your-own AI deployer: cloudflare-ai-deployer, the open-source tool behind running your own AI provider. Notebooker.ai adds the hosted service, the apps, and the polish on top of that open foundation, so you get a product that works out of the box without giving up the ability to look under the hood. Related Bring your own storage Choose your AI models Run your own AI provider ================================================================================ # Personal API keys Source: https://notebooker.ai/docs/api-keys/ Personal API keys let your own scripts and automations talk to Notebooker - save bookmarks from n8n, check whether a URL is already in your library, list notebooks, or kick off podcasts, all without a browser session. Creating a key Open Settings in the web app (user menu → Settings). In the API keys section, click Create API key and give it a name (“n8n automation”, “reading-list script”). Copy the key immediately - it starts with nbk_ and is shown exactly once. Notebooker stores only a hash of it. Keys don’t expire on their own; revoke a key from the same section whenever something no longer needs access. The list shows when each key was created and last used. Using a key Send the key as a Bearer token: curl -H "Authorization: Bearer nbk_your_key_here" \ https://app.notebooker.ai/api/v1/me The full surface lives in the API reference - every endpoint, parameter, and response schema, with a try-it console. The highlights: Endpoint What it does GET /api/v1/me Who am I - verify the key works. GET /api/v1/account Plan and remaining AI credit for the current period. GET /api/v1/sources List your saved sources. POST /api/v1/sources Save a link - body: {"url": "...", "title": "...", "notebook_ids": [...], "generate_quick_podcast": true, "podcast_format": "brief"}. Everything but url is optional. GET /api/v1/sources/check?url=… Is this URL already saved? GET /api/v1/notebooks List your notebooks. GET /api/v1/notebooks/{id}/export/markdown Export a notebook as a zip of Markdown. GET /api/v1/notes · POST /api/v1/notes List and create notes; GET /api/v1/notes/{id}/download fetches one as Markdown. POST /api/v1/search Search your sources and notes (text or vector). POST /api/v1/search/ask Ask a question, get a cited AI answer. GET /api/v1/podcasts List your podcast episodes. podcast_format accepts solo_expert, deep_dive, brief, critique, debate, or unabridged - the same format presets as the quick-podcast picker. Good to know A key acts as you: anything it saves or generates uses your account, your notebooks, and your AI credit. Treat keys like passwords. If one leaks, revoke it - revocation is immediate. Rate limits apply, the same as the browser extension. Prefer no-code? Incoming webhooks can start workflows without any key management, and watched feeds cover the “auto-import this feed” case entirely. Related API reference - the full endpoint catalog. Connect Claude with MCP - AI clients can skip keys entirely and connect over OAuth at https://app.notebooker.ai/mcp. Webhooks & workflows - trigger Notebooker from other apps. Watch RSS & YouTube feeds - hands-off ingestion. ================================================================================ # Podcast, speaker & episode profiles Source: https://notebooker.ai/docs/podcasts/ Notebooker turns the sources in a notebook into a narrated podcast episode. Two reusable profiles decide how that episode comes out: A speaker profile - who is talking: how many hosts, their names, personalities, and which voice each one uses. An episode profile - how the show is structured: the briefing/style, how many segments, the language, and which AI models write the outline and script. You pick one of each when you generate a podcast. Notebooker ships with ready-made profiles, and you can create your own. Speaker profiles A speaker profile is a named set of 1–4 speakers. Each speaker has: Field What it does Name The host’s name, used in the transcript and audio. Voice The TTS voice this speaker uses (e.g. tony, kate, melody). Backstory Background/expertise that shapes what they say (e.g. “Senior AI researcher…”). Personality Speaking style (e.g. “Analytical, warm, asks clarifying questions”). A profile also has a voice model that applies to every speaker, and each speaker can override it with their own voice. That’s how a two-host show gets two distinct voices. Built-in speaker profiles: tech_experts - two hosts (an AI researcher and a full-stack engineer). solo_expert - a single host explaining the material like a lecture. business_panel - three hosts analyzing content from a business angle. podcast_hosts - two conversational hosts for deep-dive episodes: a sharp interviewer and a well-read expert, no filler. debaters - two adversarial hosts who argue opposing positions in good faith. critic - a single frank reviewer who skips the politeness padding. Episode profiles An episode profile controls the shape and production of the show: Field What it does Speaker profile Which speaker profile (above) this episode uses. Briefing The default instructions/style for the show (e.g. “Create an engaging technical discussion…”). You can append extra guidance per episode. Segments How many segments the episode is broken into (3–10, default 5). Language The output language (e.g. en-US, pt-BR). Outline model The AI model that drafts the episode outline. Transcript model The AI model that writes the spoken script. Built-in episode profiles: tech_discussion - pairs with tech_experts, 5 segments, technical tone. solo_expert - pairs with solo_expert, 4 segments, explanatory tone. business_analysis - pairs with business_panel, 6 segments, business tone. Format presets - episode profiles shaped after the ways people actually want to listen. They’re also the choices in the quick-podcast format picker when you save a link: deep_dive - two hosts (podcast_hosts) work through all of the material, citing specifics from your sources. 8 segments, no banter, no repeated recaps. brief - a single voice delivers just the essentials, most important takeaway first. 3 segments, no preamble. critique - the critic reviews your sources like a frank peer reviewer: weak claims, missing evidence, counterarguments - and what genuinely holds up. debate - the debaters steelman the strongest case for and against the material’s central claims, with direct rebuttals. unabridged - an exhaustive section-by-section walkthrough for a listener who will never read the original. 10 segments; completeness over speed. How they fit together Episode profile ──references──▶ Speaker profile • briefing/style • speaker 1 (name, voice, personality) • segments • speaker 2 … • language • (1–4 speakers) • outline + transcript models When you generate an episode, Notebooker: Reads your episode profile for the briefing, segment count, language, and which models to use. Looks up the speaker profile it references to get each host and voice. Writes an outline, then a full script with each line attributed to a speaker. Synthesizes each line in that speaker’s voice and stitches the audio together. Publishes the finished episode to your library and your private podcast feed. Creating and editing profiles Profiles live in your Notebooker settings. You can: Duplicate a built-in profile and tweak it - the fastest way to get a custom show. For example, duplicate tech_experts, rename a host, and change their voice to belinda. Create a profile from scratch, defining each speaker and the episode briefing. Edit or delete any profile you own. Tip: keep the number of speakers in the speaker profile consistent with the conversational style in your episode briefing. A “panel debate” briefing with a single-speaker profile will read like a monologue. Related Notebooker TTS voices - the voices you assign to speakers. Subscribe to your podcast feed - listen on your phone. ================================================================================ # Publishing & embeds Source: https://notebooker.ai/docs/websites/ Every notebook can become something you share, not just something you read. Notebooker publishes two kinds of things: Websites - the Websites creator builds a real multi-page static site from a notebook, and publishing puts the whole site live. Any other creation with its own viewer - flashcards, charts, infographics, mindmaps, timelines, wikis, Socratic sessions, counterpoint debates - publishes as a single self-contained page: the same interactive viewer you use inside Notebooker, at a public link anyone can open. No account needed to view it. Either way you also get an embed snippet - one <iframe> tag to paste into a blog post, course page, Notion doc, or anywhere else HTML goes. The Websites creator The Websites creator turns a notebook into a real, multi-page static website: a home page that welcomes a first-time visitor and points into the rest of the site, plus a page for each major topic in your sources - cross-linked, with a navigation bar, an optional sidebar listing every page, and built-in search. Everything on the site is written only from your material. Notebooker first plans the site like an information architect - which pages, covering what, drawn from which sources - then writes each page and renders the whole thing with Quarto, the same engine behind quarto.org itself and course.fast.ai. The result reads like a credible reference site, not a template blog. Creating a website From a notebook, choose Website in the creators list and set your options: Pages (2–15, default 6) - includes the home page. Theme - ten Bootswatch looks, from the clean default cosmo to dark themes like darkly and cyborg. Search - Quarto’s built-in client-side site search, on by default. Sidebar - a docked sidebar listing every page, on by default. As with every creator, your instructions steer the result: audience, tone, what to emphasize, what to leave out. Downloading the site Every website artifact includes a website.zip - the complete rendered site, ready to host anywhere static files are served: GitHub Pages, Cloudflare Pages, Netlify, an S3 bucket, or just opened from disk. Unzip it and index.html is the front door; there’s no build step and nothing to install. Publishing - websites and everything else If you’ve connected your own storage (any S3-compatible bucket - Cloudflare R2, DigitalOcean Spaces, AWS S3) with a public URL configured, the publish action on a finished creation puts it on the web and hands you the link. A website publishes as the full rendered site - every page, style, and the search index. A flashcard deck, chart, mindmap, timeline, wiki, Socratic session, or debate publishes as one self-contained page carrying its interactive viewer - a flashcard link is a studyable deck, not a screenshot. A small “Made with Notebooker” footer credits the source. Everything uploads to your bucket under your own account prefix - it lives in your storage, on your domain, and stays up as long as you keep it there. Publishing again after regenerating replaces the page at the same link; deleting the creation removes the published files too. A few notes: The public URL comes from your storage settings (for R2 that’s a custom domain or the bucket’s r2.dev URL; for Spaces, the CDN endpoint). If no public URL is configured, Notebooker will tell you before anything is uploaded. Anything you publish is publicly readable at that link - publish only what you’re comfortable sharing. Creations without their own viewer (for example essays, which are documents you download) aren’t publishable this way - share the PDF instead. Embedding Every publish also gives you a copy-embed button with a ready-to-paste snippet: <iframe src="https://your-domain/.../index.html" width="100%" height="640" style="border:0;border-radius:8px" loading="lazy"></iframe> Paste it into anything that accepts HTML - a course page, a blog post, internal docs - and the live, interactive creation renders inline. Adjust height to taste; the page is responsive inside the frame. Tips Name your audience in the instructions. “A reference site for new team members” and “a public explainer for customers” produce very different pages from the same sources. Fewer, deeper pages usually beat many thin ones. Six pages is a sweet spot for most notebooks; go higher only when your sources genuinely cover that many distinct topics. Try a dark theme for docs you’ll read at night - darkly and cyborg are full dark modes, not just accents. Embed decks where the learning happens. A published flashcard deck embedded in a course page means students study in place - no sign-ups, no links out. ================================================================================ # Run your own AI provider Source: https://notebooker.ai/docs/custom-ai-provider/ Notebooker can use any OpenAI-compatible endpoint as an AI provider. If you’d like to run the models yourself - for cost control, data residency, or just to tinker - we make it easy to deploy one to your own Cloudflare account, backed by Cloudflare Workers AI. The free plan includes a daily allowance of inference, so you can try everything without a card. There are two ways to do it: the hosted deployer at open.notebooker.ai (nothing to install), or running the same open-source app locally. Both deploy the exact same worker. What it deploys A Cloudflare Worker in your account that exposes an OpenAI-compatible REST API: GET /v1/models list configured models POST /v1/chat/completions chat / language model (incl. vision models) POST /v1/embeddings embeddings POST /v1/audio/transcriptions speech-to-text POST /v1/audio/speech text-to-speech Each endpoint is enabled only if you’ve configured a model for it. Option 1 - open.notebooker.ai (nothing to install) Open the deployer at open.notebooker.ai → Prepare your Cloudflare account (one time): verify your account email, and visit Workers AI in your dashboard once so your workers.dev subdomain gets registered. Create a scoped API token at dash.cloudflare.com → API Tokens (Custom Token, scoped to your account): Workers Scripts: Edit, Workers KV Storage: Edit, Workers AI: Read, Account Analytics: Read. Paste the token at open.notebooker.ai, pick a model for each slot from the live Cloudflare catalog, and hit deploy. Test it right there - streaming chat, text-to-speech, speech-to-text (upload audio/video), vision (upload an image), and embeddings similarity - and watch your estimated free-tier usage. Save your credentials: copy the base URL and bearer key, or click download credentials.txt. You’ll paste these into Notebooker below. A few things worth knowing: Nothing is stored on our side. Your Cloudflare token lives only in an encrypted cookie that expires with your session - delete the token in Cloudflare when you’re done, and create a new one next time. Your endpoint’s API key is never stored either: save it when it’s shown (or download the credentials file). If you come back without it, paste the saved key in, or renew it (which replaces the old one). Model changes are safe - redeploying with different models never touches your API key. Gated models: some Meta models (e.g. llama-3.2-11b-vision-instruct) require a one-time license agreement per Cloudflare account. If chat replies with a license notice, send the single message agree once and you’re set. Option 2 - run the deployer locally The deployer is open source - cloudflare-ai-deployer and the hosted app is the same code, so you can run it on your own machine (Node.js 18+): git clone https://github.com/Notebooker-ai/cloudflare-ai-deployer cd cloudflare-ai-deployer/web npm install npm run dev # http://localhost:4321 Then follow the same flow as Option 1 on your local instance: paste your scoped token, pick models, deploy, test, and download your credentials. See the repo README for the architecture and how to deploy your own hosted instance. Point Notebooker at it In Notebooker: Go to Settings → API keys and Add credential → OpenAI Compatible. Fill in: Name - e.g. “My Cloudflare AI” API key - your endpoint’s bearer key Base URL - e.g. https://cloudflare-ai.<your-subdomain>.workers.dev/v1 Save. Go to Settings → Models, Add model, choose provider OpenAI Compatible, and enter the model name matching a configured slot (e.g. chat, embedding, text_to_speech, speech_to_text). Now select your model wherever Notebooker asks which model to use - for chat, embeddings, podcasts, and creators. Requests to your own endpoint run on your Cloudflare account’s Workers AI allowance, not your Notebooker credit - Notebooker only counts Notebooker AI and service usage. Verify the endpoint any time with: curl https://cloudflare-ai.<your-subdomain>.workers.dev/v1/models \ -H "Authorization: Bearer <your-api-key>" Related Notebooker TTS voices - using the speech model. cloudflare-ai-deployer on GitHub ================================================================================ # Search, Ask & Chat Source: https://notebooker.ai/docs/search/ Once you’ve saved a few sources, there are three ways to get something back out of them, depending on how sure you already are of what you’re looking for. Full-text search Search runs across everything in your library at once: source text, notes, and the insights Notebooker has generated. It is the fastest way to jump to a passage when you already know roughly what you’re after. Matches link straight to the place they appear so you can read them in context. Ask Ask takes a question and answers it over your whole notebook. Notebooker retrieves the most relevant material on its own, writes a grounded answer, and cites every claim back to the source it came from. A coverage strip shows how many of the documents it consulted actually made it into the answer, so you can tell a thorough answer from a thin one. Use Ask when you don’t know which document holds the answer, or when the answer lives across several of them. Chat Chat works over a context you select. Pick the sources you care about and the model reasons over just those, in full. Because the context is smaller and deliberate, Chat is the tool for close reading, follow-up questions, and drafting from a specific set of documents. A rule of thumb: Ask searches everything and finds what’s relevant; Chat stays inside the sources you chose. Related Citations & answer coverage How Notebooker handles content Insight personas ================================================================================ # Socratic tutoring Source: https://notebooker.ai/docs/socratic/ The Socratic tutor creator turns a notebook into a tutoring session built on one rule: it never blurts the answer. You reason your way to each concept through questions, and the session only reveals a model answer after you’ve worked for it - or never, if you turn reveals off. It exists because “act as a Socratic tutor” prompts share a famous weakness: ask a chatbot twice and it caves and explains. Notebooker sidesteps that structurally. The whole dialogue is written before you start - from your sources, with citations - and the session simply doesn’t put an answer on screen until the flow reaches it. How a session flows One question at a time. Each concept (“nugget”) opens with a single question - sometimes clarifying, sometimes a counterexample, sometimes a what-if. A progress ledger at the top fills in as you go. Answer in your own words. There’s a scratch box to think out loud in; writing the reasoning is the point. Check yourself. Press Check myself and grade your answer against a short checklist of the points a correct answer would cover. You do the grading - the session trusts you, and the summary reflects what you ticked. Wrong turn? Get a question back. Tempting misconceptions appear as “I think…” buttons. Tap one and you get a counter-question that exposes the contradiction - never a correction. Stuck? Ask it smaller. I’m not sure - ask me smaller gives up to three hints, each a smaller question that breaks the problem down. The last one points right at the key idea - but still asks. The reveal, last. Once hints are exhausted, the model answer unlocks on the check screen, cited to the exact sources it rests on. Nail a concept fully and you earn an optional stretch question instead. Finish with a map, not a grade. The summary sorts concepts into demonstrated, revisit, and needed the answer - so you know exactly what to reread. Copy the report or run the session again. At synthesis difficulty, the final questions are capstones: they unlock only after every concept is done, and they can’t be answered without combining ideas from across your sources. Options Option What it does Number of concepts 3–25 nuggets (default 10) - a 10-nugget session is roughly a 20–40 minute sitting Difficulty recall (definitions and facts), application (reasoning with the material), synthesis (adds capstone questions) Tutor persona The voice of the session - “a patient, curious professor” by default; try “a drill sergeant” or “a friendly peer” Show model answers On by default. Off = pure Socratic: the session never states an answer, ever Scope A boundary like “only material through chapter 5” - later content is treated as spoilers and left out entirely Grounded in your sources Every question, hint, and reveal is generated from your notebook - not the model’s general knowledge - and reveals cite the specific sources they rest on. If the model tries to cite a source that isn’t in your notebook, the citation is dropped rather than shown. The scope option is made for textbooks and fiction: cap the session at a chapter and nothing past it leaks into the questions. The study sheet Every session also produces a downloadable Markdown study sheet laid out like a good problem-set book: questions and collapsible hints up front, model answers and checklists in an answer section at the back. Print it, or drop it into your notes app and quiz yourself away from the screen. Socratic tutor, flashcards, or study guide? Socratic tutor teaches reasoning: work through ideas step by step, with your misconceptions challenged along the way. Best the first time you’re learning something, or before an exam or interview that will make you explain. Flashcards train recall on a schedule - spaced repetition keeps facts fresh over weeks. Study guides are for reading: a topic-by-topic reference you review rather than interact with. They stack well: run a Socratic session first, then make flashcards for whatever landed in the revisit column. ================================================================================ # Subscribe to your podcast feed Source: https://notebooker.ai/docs/podcast-feed/ Every podcast you generate is added to a private RSS feed unique to your account. Add that feed to any podcast app and your episodes show up automatically, ready to download and listen to offline - just like any other show. Get your feed URL Open Notebooker and go to your Library. Filter to Podcasts. Find the “Subscribe in any podcast app” card and copy the feed URL. Your feed URL looks like: https://app.notebooker.ai/podcasts/feed/<your-private-token> The token in the URL is what keeps your feed private - anyone with the link can see your episode titles and listen, so treat it like a password. If it ever leaks, use Regenerate on the same card to issue a new URL (this breaks any existing subscriptions, so you’ll need to re-add the new link in your podcast app). Add it to a podcast app Most podcast apps support adding a show “by URL” (sometimes called “add by RSS” or “add a private feed”). Paste your feed URL when prompted. App How to add a feed by URL Apple Podcasts (iOS / macOS) Library → top-right menu → Add a Show by URL… Pocket Casts Profile → Add a podcast → search box → paste URL Overcast + (top right) → Add URL AntennaPod (Android) Add Podcast → Add Podcast by RSS address Castro Add (+) → paste URL Podcast Addict (Android) + → RSS / podcast feed → paste URL gPodder / desktop readers “Add podcast” → paste URL Apps that don’t expose an “add by URL” option (notably Spotify and some in-car/voice assistants, which only list shows from their own directory) can’t subscribe to a private feed. Use one of the apps above for those. Tip: Your feed only includes episodes that have finished generating and have audio. A podcast you just kicked off will appear in the feed once it’s ready - pull-to-refresh in your podcast app to fetch it. What’s in the feed The feed is standard RSS 2.0 with iTunes podcast tags, so it works anywhere: One entry per finished episode, newest first. Each entry has the episode title, its briefing as the description, a publish date, and the downloadable audio. Episodes still generating are left out until their audio is ready. Related Podcast, speaker & episode profiles - shape the episodes that land in this feed. Notebooker TTS voices - choose how your hosts sound. ================================================================================ # Usage, credit & budgets Source: https://notebooker.ai/docs/usage/ Notebooker bills for what you actually use, and shows you the cost as you go. Your plan includes a monthly credit; anything past it is pay-as-you-go, capped by a budget you control. What the credit covers Your monthly credit goes toward both the AI and the service work behind each action, including: AI usage - the models that chat, generate insights, and narrate podcasts. Fetching and archiving sources - snapshotting pages, downloading media, and extracting text. Storage past your included amount - keeping your sources, audio, and creator artifacts. Your first 5 GB is free; only bytes above that are billed, at $0.10 per GB-month, prorated daily. Or skip Notebooker storage for new files entirely - bring your own bucket and migrate what’s already stored. Every action shows its cost, so nothing is hidden in a bundle. You can see a running total of what you’ve used on your Dashboard. When the credit runs out Heavy months roll into pay-as-you-go overage, and it’s capped by a budget you set. That budget defaults to $0, so you never spend more than your plan without deliberately opting in. Raise the cap when you want the extra headroom, or leave it and keep collecting until the next cycle. Bring your own keys Connect your own AI keys and those requests are billed by your provider directly, not against your Notebooker credit - Notebooker only counts Notebooker AI and service usage (fetching and archiving sources, and Notebooker-managed storage). Pair that with your own storage and the largest usage costs move onto infrastructure you already pay for. Related Choose your AI models Bring your own storage Pricing ================================================================================ # Watch RSS & YouTube feeds Source: https://notebooker.ai/docs/watched-feeds/ The Watch Feed workflow action keeps a feed and your library in sync: every time it runs, it fetches the feed, figures out which entries it hasn’t seen before, and saves each new one as a link source - optionally straight into a notebook. Pair it with a Schedule trigger and your reading list fills itself. Set it up Go to Workflows and create a workflow. Choose the Schedule trigger and pick how often to check (hourly and daily are typical; feeds rarely need more). Add the Watch Feed action and configure it: Setting What it does Feed URL Any RSS 2.0 or Atom feed. Notebook ID (optional) New sources are added to this notebook. Max new items per run Caps each run (1–50, default 10). The first run imports up to this many of the latest entries; later runs import only what’s new. Watching a YouTube channel Every YouTube channel has a hidden Atom feed: https://www.youtube.com/feeds/videos.xml?channel_id=CHANNEL_ID Find the channel ID on the channel’s About page (or in the page source as channelId), drop the URL into Feed URL, and new uploads arrive as video sources - transcribed and searchable like anything else you save. How it behaves No duplicates. Notebooker remembers which entries it has imported for each feed, so re-runs and overlapping schedules are safe. Bulk-lane processing. Feed imports run at low priority - they never queue ahead of things you save by hand, and they skip the quick-podcast option. Credit-aware. Each imported link uses AI credit like any other source. If your credit runs out mid-run, the action stops and reports how far it got; unimported entries stay unseen and import on a later run once credit is back. Auto-snapshot applies if you’ve enabled it in Settings, same as any other new source. Related Webhooks & workflows - the workflow engine this action lives in. Importing & bulk actions - one-time bulk imports of bookmark files. Personal API keys - script your own ingestion instead. ================================================================================ # Webhooks & workflows Source: https://notebooker.ai/docs/webhooks/ Notebooker connects to the rest of your tools through Workflows. A workflow has a trigger (what starts it) and one or more actions (what it does). Webhooks show up on both sides: Incoming webhook - an outside app sends Notebooker a request to start a workflow. Outgoing webhook - a workflow sends a request to an outside app as one of its actions. You build both under Workflows in Notebooker. No code required. Incoming webhooks (start a workflow) Use an incoming webhook when another service should kick off a Notebooker workflow - for example, “when my form is submitted, save the link as a source” or “when CI finishes, generate a podcast summary.” Set it up: Create a workflow and choose the Incoming Webhook trigger. Notebooker generates a unique URL with a secret token in it: https://app.notebooker.ai/api/webhooks/workflows/<token> Configure the other service to POST JSON to that URL. Send a request: curl -X POST https://app.notebooker.ai/api/webhooks/workflows/<token> \ -H "Content-Type: application/json" \ -d '{"url": "https://example.com/article", "title": "Something to read"}' Notebooker accepts the request (HTTP 202) and runs every active workflow that uses that token. Use the data: any field you send is available to your workflow’s actions as a variable - for example and. So a “Create Source” action can save exactly the link that was posted in. The token in the URL is the credential - keep it secret. If it leaks, regenerate it in the trigger settings. Outgoing webhooks (notify another app) Use an outgoing webhook when a workflow should tell another service something happened. Add an HTTP Request action to any workflow. Configure the action: Field What it does Method GET, POST, PUT, PATCH, or DELETE. URL The endpoint to call - supports variables like ``. Headers Any custom headers (e.g. an Authorization token for the receiving service). Body JSON or form data, also with variables, for POST/PUT/PATCH. Timeout / retries How long to wait and how many times to retry on failure. Store response Optionally capture the response for later actions in the workflow. For example, a workflow triggered by “Podcast created” can POST to a Slack or Discord incoming-webhook URL so your team gets a message every time a new episode is ready. Triggering on Notebooker events Beyond incoming webhooks, workflows can also start when something happens inside Notebooker: Source created / updated / deleted Notebook created / updated / deleted Podcast created / updated / deleted Pair one of these with an HTTP Request action and you’ve built an outgoing event webhook - “whenever a podcast is created, notify my app.” What a workflow can do Besides HTTP Request, workflow actions include: Create source, Create notebook, Add source to notebook - build your library automatically. Watch Feed - import new entries from an RSS/Atom feed or YouTube channel on a schedule. Create podcast - generate an episode as part of a flow. Send email - email yourself or a teammate. Condition and Run workflow - branch on data and chain workflows together. Services that work well Anything that can send or receive a JSON HTTP request integrates cleanly - Slack, Discord, Zapier, Make, n8n, IFTTT, GitHub, Linear, Notion, Telegram, and more. See the Integrations page for a fuller list and links to each service’s webhook docs. Note on billing webhooks Notebooker also receives Stripe webhooks to keep your subscription and credits in sync. That’s an internal, system-level integration handled for you - it isn’t something you configure, and it’s separate from the Workflow webhooks above.