AI note analysis
An LLM classifies each note into Tasks, Deadlines, Schedule, Thoughts or Learning, detects priority (Low/Medium/High), writes a title and extracts natural-language dates like “tomorrow at 3pm”.
SmartNote Scheduler is an AI-powered note app. Write in plain English — it reads your note, classifies it, sets the priority, pulls out the date and time, and reminds you before it's due.
Five auto-detected categories — Tasks, Deadlines, Schedule, Thoughts and Learning — with no folders to file.
Everything below is implemented in the current release — no roadmap promises dressed up as features.
An LLM classifies each note into Tasks, Deadlines, Schedule, Thoughts or Learning, detects priority (Low/Medium/High), writes a title and extracts natural-language dates like “tomorrow at 3pm”.
Reminders fire one hour before a deadline or scheduled time, driven by a durable heartbeat endpoint at POST /api/scheduled/process-reminders.
Push notifications plus optional email through a provider-neutral webhook. Every attempt is logged as sent, failed or skipped — nothing fails silently.
Full-text search across titles and content, category tabs, a “hide completed” toggle, and sorting by date, priority or category.
Totals, completed, pending and overdue counts, with per-category summaries and overdue highlighting so the slipping items are obvious.
JSON backup export and import, plus CSV and Markdown exports and reusable filter presets. Your notes stay portable.
No forms, no date pickers, no category dropdowns. Write the sentence you'd say out loud.
Type a note the way you'd say it — “renew the insurance by Friday” or “standup tomorrow at 9am”.
The LLM categorises the note, sets a priority, generates a title and extracts the date and time in the same pass.
A reminder is scheduled one hour ahead and delivered by push or email — logged every time it fires.
An illustrative reconstruction of the notes dashboard: category tabs, priority pills and the overdue-first ordering.
The screenshot above is a stylised mock, not a live capture — the repository does not publish a hosted demo URL, so this is how the real dashboard is laid out.
SmartNote Scheduler is a Node + React application, not a static page — it needs a MySQL-compatible database and a few environment variables.
pnpmDATABASE_URL, JWT_SECRET, OAUTH_SERVER_URL and VITE_APP_ID in a .envInstall dependencies, push the Drizzle migrations, and start the dev server. Email delivery is optional — leave EMAIL_WEBHOOK_URL unset and deliveries are recorded as skipped instead of failing.
# clone git clone https://github.com/drsamonline/SmartNotes.git cd SmartNotes # install pnpm install # configure .env, then migrate the database pnpm db:push # development (server + Vite HMR) pnpm dev # tests / type check / production build pnpm test pnpm check pnpm build && pnpm start
It reads a note you wrote in plain language and, in one pass, categorises it into Tasks, Deadlines, Schedule, Thoughts or Learning, detects a priority of Low, Medium or High, generates a title, and extracts any date or time such as “tomorrow at 3pm”.
Reminders fire one hour before a deadline or scheduled time. Delivery is by push notification and, optionally, email through a provider-neutral webhook. Each delivery attempt is recorded as sent, failed or skipped in the notification history.
Yes. SmartNote Scheduler stores notes, reminders and notification logs in a MySQL-compatible database via Drizzle ORM, and ships schema migrations you apply with pnpm db:push.
Reminders are processed by an external scheduler that calls POST /api/scheduled/process-reminders. After deployment you wire up a one-minute heartbeat job pointing at that endpoint. Reminders that only failed or were skipped stay pending and are retried once an email provider is configured.
Yes — JSON backup export and import, plus CSV and Markdown exports, along with reusable filter presets.
MIT. The source, changelog, architecture notes and API reference all live in the repository.