NoticeBridge

School circulars, finally clear. NoticeBridge is a family checklist that catches changed deadlines and links every task to the original notice.

Comment

Inspiration

The important detail is often in the second school message.

The first circular says the consent form is due on 24 October. The revised circular says 27 October and adds a school ID card to the packing list. Families have to notice both changes while keeping track of everything they have already done.

NoticeBridge grew from that everyday problem: helping students and parents turn school messages into a clear, dependable plan.

What it does

NoticeBridge is a personal school noticeboard. Upload a PDF or English screenshot, or paste a circular, and it turns explicit instructions into a checklist.

  • Check every instruction: Each task links to the exact words supporting it in the circular.
  • Catch revised plans: Compare two versions to identify changed deadlines, updated instructions, additions, and removals.
  • Keep your progress: Unchanged instructions retain their completion state. Changed instructions need checking again.
  • Confirm dates before exporting: Review deadlines against the source, then export confirmed, unfinished tasks to a calendar file.
  • Share the plan: Copy, save, or print a family checklist containing instructions, dates, and source quotes.
  • Keep your board: Personalize it with a name, school, and class, then back up and restore circulars and progress.

The interface uses the visual language of a school noticeboard: paper sheets, pinned deadlines, handwritten notes, and red revision marks.

How it was built

The application uses Next.js, React, and TypeScript. PDF.js extracts PDF text, Tesseract.js handles English OCR, and a local extraction engine identifies instructions and dates. Zod validates application data and imported backups.

The default workflow processes notices in the browser and saves the board on the device. Browser speech synthesis provides read-aloud functionality. An optional OpenAI integration supports richer extraction and Hindi translation when configured.

The project has 46 passing automated tests covering extraction, dates, revision matching, calendar exports, backups, and API behavior. The production build also passes.

Challenges

Keeping evidence attached to instructions. A shortened task can lose important context. Each extracted task therefore retains an exact source quote that users can inspect.

Handling uncertain dates. Missing years and ambiguous wording should not silently become calendar reminders. Dates need confirmation before export.

Comparing revisions without erasing progress. A new circular may change one deadline while leaving other instructions untouched. Revision matching preserves unchanged progress and resets changed tasks for review.

What we learned

A useful summary needs a way back to the original wording. Showing the evidence makes checking an instruction straightforward.

Revisions also need more than a new checklist. Families need to see what changed, what still applies, and which completed steps remain valid.

These lessons shaped the central workflow: extract, inspect, compare, and confirm.

Current scope and what comes next

The demonstrated prototype uses local English extraction and English OCR. English/Hindi board controls and Hindi sample instructions are available; full translation of imported notices requires the optional AI configuration.

Calendar files are imported manually, and boards use local storage with file backups rather than cloud synchronization. The demo uses a fictional school circular and does not use live AI processing.

Further work would focus on broader language coverage, more varied circular formats, and optional shared family boards.

Tools and credits

Built with Next.js, React, TypeScript, PDF.js, Tesseract.js, Zod, Lucide, Vitest, browser speech synthesis, localStorage, and iCalendar exports.

Typography uses Google Fonts: Barlow Condensed, IBM Plex Sans, IBM Plex Mono, and Caveat. The Greenwood Public School circular is fictional demonstration content.

The narrated demo uses actual browser captures assembled with Python and FFmpeg, with macOS speech synthesis for narration. Optional OpenAI provider tests use mocked responses.

Built With

Share this project:

Updates

Submission history