Study, organize and automate inside a single Project

What Esmerum is

Esmerum is a study platform that gathers, inside a single Project, everything that normally lives across five different apps: the files and PDFs, the notes, the visual board of ideas, the classroom, the spaced-repetition flashcards and the automations that tie one to another.

The practical difference is shared context. What you read in Notebook becomes a question in Questions without leaving the app, and the AI assistant answers by looking at the whole Project — not at a file you just pasted into a chat.

Five workspaces, one context

Every Project opens with the same five workspaces. They are not separate modules: they share the same library, the same assistant and the same automations.

• Notebook — block-based document editor with markdown, tables, embeds, references and backlinks.
• Canvas — infinite visual board for sticky notes, text and connections between ideas.
• Classroom — modules, lessons, materials, schedule and class progress tracking.
• Questions — decks, topics and subtopics, with flashcards, multiple choice and true/false governed by FSRS.
• Automation — visual flows of connected nodes, with event triggers, schedules and MCP connectors.

How Esmerum picks the next question

FSRS-4.5 decides when to review. A rule-based selector decides what: memory urgency, recent and lifetime error rate, topic mastery, difficulty fit and how long it has been since that content last appeared.

Because the choice is a rule and not a model, every question can explain why it showed up now — "you missed 2 of the last 3 in this topic, and mastery sits at 46%" — with no second AI call and no extra second of waiting.

• Four review grades: again, hard, good and easy.
• The same question changes shape as memory strengthens: multiple choice, cloze, true/false and finally flashcard.
• Reloading mid-session gives back the same queue; coming back later reshuffles the top.
• Timed mock exams freeze their question set, grade at the end, and leave the spaced-repetition schedule untouched.

Automations that study alongside you

Automations are visual flows of connected nodes, in the spirit of n8n, and you can either drag the nodes yourself or describe in one sentence what you want — the agent builds the flow.

Missing a review can generate a batch of new questions in the same topic without you opening anything. Cost guards run before the AI call: how many questions you already have open, and how much depth that topic has already gained.

• Event and schedule triggers: country, timezone, time of day and weekdays, with no cron expression to write.
• MCP connectors to notify over WhatsApp or Google Calendar, with the token encrypted on the server.
• Filter, AI and action nodes, chainable across the workspaces of the same Project.

An assistant that shows where the answer came from

The side chat searches the Project library whenever the question has a subject, and only opens Notebook when you ask. Every answer leads with a badge for what it consulted, and the citations inside the text are clickable: each one opens the exact passage that supported that sentence, with its position inside the document.

When no passage in the library supports the answer, the model answers without citing rather than forcing a source that is not there. A link that does not reach the source lies more than text with no link.

In the browser, on the desktop and in your pocket

The web version runs in any modern browser. The desktop app runs on Tauri — a native window, keyboard shortcuts and the same Esmerum without browser tabs — with builds for Windows, macOS and Linux. On Android, reviews and the assistant fit in your hand, and the app doubles as a remote control for the desktop session.

Studying is local-first: reviews are written on the device and synced afterwards, so losing connection in the middle of a session interrupts nothing.

Frequently asked questions

What is Esmerum?

Esmerum is a study platform built around Projects: files, notes, a visual board, spaced-repetition flashcards, classrooms and visual automations, all with integrated AI, in one place.

Do I have to pay to start?

No. The Free plan gives you all five workspaces with usage limits and asks for no credit card. You only change plans when the limit starts to get in the way.

Which spaced-repetition algorithm does Esmerum use?

FSRS-4.5, with four review grades (again, hard, good and easy). The algorithm decides when to review; a rule-based selector decides which question appears, weighing memory urgency, error rate, topic mastery, difficulty and time since the last exposure.

Does the AI write the questions on its own?

It does, from your own material, and you can still write and edit everything by hand. Generated questions pass through a rule-based audit that repositions the correct answer, drops duplicated options and demotes to a flashcard any item whose shape gave the answer away.

How does the Marketplace work?

Creators and teachers publish study packages — video courses, flashcard decks and full libraries. When you get a package it connects straight into your Esmerum and keeps receiving the creator updates, with no downloads and no stray copies.

Does it work offline and on the desktop?

The desktop app (Windows, macOS and Linux) brings the full experience to a native window. Studying is local-first: reviews stay on the device and upload when the connection returns.

Which languages is Esmerum available in?

Six: English, Portuguese, Spanish, French, Italian and German. The language is detected from the browser, with English as the default when none of the six is recognized.

Can I use my own AI API key?

Yes, on any plan. Anything that runs through your key does not consume the AI quota included in the plan.

Esmerum in numbers

The statements below describe how the platform works and can be quoted directly.

• Esmerum organizes studying into Projects, and every Project opens with five workspaces: Notebook, Canvas, Classroom, Questions and Automation.
• Spaced repetition runs on the FSRS-4.5 algorithm, with four review grades: again, hard, good and easy.
• Which question comes next is decided by a rule, not by a model: memory urgency, error rate, topic mastery, difficulty fit and time since the last exposure.
• Every question states why it appeared at that moment, without costing a second AI call.
• The same question changes shape as memory strengthens: multiple choice, cloze, true/false and finally flashcard.
• The interface ships in six languages — English, Portuguese, Spanish, French, Italian and German — detected from the browser.
• The free plan covers all five workspaces with usage limits; Pro costs $9.90 per month and delivers 12× the free AI quota.
• Ultra costs $99 per month and delivers 10× the Pro quota, keeping the same ratio between price and usage.
• Annual billing charges twelve months at the price of nine, a 25% discount.
• Currency follows the reader: Brazilian customers pay in reais through Pix (Pro at R$ 49, Ultra at R$ 490); everyone else pays in US dollars by card subscription.
• Every plan lets you connect your own API key, and anything that runs through it does not consume the Esmerum quota.
• Automations are visual flows of connected nodes, with event and schedule triggers, and no cron expression to write.
• MCP connectors send notifications over WhatsApp, Google Calendar and other services; the token stays encrypted on the server and never inside the flow.
• The assistant answers from the material in your own Project and shows its sources: every citation opens the exact passage that supported the sentence.
• The desktop app runs on Tauri, with builds for Windows, macOS and Linux; Android ships as an APK, and iOS is planned.
• Studying is local-first: reviews are written on the device and synced afterwards, so losing connection does not interrupt a session.
• Mock exams freeze their question set at creation, are timed by the server, and never alter the spaced-repetition schedule.