Skip to content
Devs-GMTPublic

About

AI notebook that turns your notes or a short PDF into a summary, key points, a quiz and answers, then stamps your study proof (title, notes hash, score) on BOT Chain.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

PocketLM mascot

PocketLM

An AI notebook that actually helps you learn, then proves it.

Paste text or a short PDF and get a summary, 5 key points, a 5-question quiz and answers to your questions. Listen to the summary, finish the quiz, and stamp your study proof (title, a fingerprint of your notes and your score) on BOT Chain. Your notes never leave your browser except to be summarized, and they are never on-chain.

Live App | Contract (mainnet) | Source | How it works | Screens | Run it locally | Limits

Solidity Hardhat tests Next.js Groq BOT Chain No accounts License: MIT

PocketLM landing page

Screenshots

Home Study kit summary
PocketLM home Summary tab
Drop a PDF or paste notes. Recent notebooks are saved only in your browser. One request makes the summary, 5 key points and a quiz, with the source details beside it.
Quiz Ask your notes
Quiz tab Ask tab
One question at a time. Right or wrong shows straight away, with the reason. Answers come only from your document and quote the line they used.
Landing: how it works
How it works section
Four pages, one notebook: add notes, get a study kit, ask and quiz yourself, then stamp your study on BOT Chain.

How it works

  1. Paste text or upload a PDF (up to 10 pages / ~20,000 characters). PDF text is read in your browser with pdf.js.
  2. The text goes to our own API route (/api/study), which asks Groq for a summary, 5 key points and a 5-question quiz. /api/ask answers questions only from your document and quotes the line it used.
  3. Listen uses the browser's built-in voice (sentence by sentence, with speed control). Nothing extra is sent anywhere.
  4. The quiz is graded in your browser, one question at a time with the reason after each answer. Finish it and stamp title + SHA-256 of the text + score on BOT Chain from a confirmation dialog that shows exactly what goes on-chain.
  5. Study records and /app?study=ID read the proof straight from the contract. No database, no accounts.

Only the title, the document hash and the score are on-chain. The document text never is.

Screens

Route What it is
/ Landing page (hero video with a crossfade loop, how it works, features, FAQ, Pocket the guide)
/app Home: drop a PDF or paste notes, recent notebooks
/app/new Create a notebook (PDF or pasted text)
/app/notebook?id= Summary, Key points, Quiz, Ask, Listen
/app/notebooks My notebooks: search and delete
/app/records Study records read from the contract, with stats and a time filter (/shelf redirects here)
/app/settings Wallet, network and contract, delete local data

Notebooks (your notes, the study kit and quiz progress) are saved only in your browser (localStorage, up to 30). Clearing browser data removes them; anything already stamped on-chain stays.

Contract

contracts/src/PocketLM.sol (Solidity 0.8.24, no tokens, no payments, no admin).

Network Chain ID Address
BOT Chain Testnet (development only) 968 0x9e7ad905539f72c3Ae7B8f8637c19EEFBa8B61aA (scan.bohr.life)
BOT Chain Mainnet 677 0xa0AD40C50001DEFcEe354Ec0316b99F78bD6c64f (scan.botchain.ai)

Study ids start at 1. Titles are limited to 80 bytes. score <= total and total > 0 are enforced.

Project layout

PocketLM/
  contracts/   Hardhat project: src/PocketLM.sol, test/, scripts/deploy.js, deployments/
  web/         Next.js app: pages and UI (app/, components/), browser logic and the
               Groq proxy (app/api/*, lib/), deployed to Vercel

Run it locally

npm run install:all                               # installs web/ and contracts/
cp contracts/.env.example contracts/.env          # deployer key, testnet only
cp web/.env.local.example web/.env.local          # GROQ_API_KEY
npm run dev                                       # web app on http://localhost:3000
npm run test:contract                             # contract tests
npm run deploy:testnet                            # deploys to BOT Chain testnet, writes contracts/deployments/botTestnet.json

Secrets live only in contracts/.env and web/.env.local (both git-ignored).

Deploy

The app is live at https://pocketlm.my.id/ (BOT Chain mainnet). On Vercel, set the project's Root Directory to web and add these Environment Variables. The deployer key is never needed there.

Variable Value
GROQ_API_KEY your Groq key (secret)
GROQ_MODEL openai/gpt-oss-120b (optional)
NEXT_PUBLIC_NETWORK optional, mainnet is the default; set testnet only for development

The link preview image and page metadata use the production URL that Vercel provides. With a custom domain (the live app uses pocketlm.my.id), set NEXT_PUBLIC_SITE_URL=https://pocketlm.my.id so the preview links point to it.

Mainnet (done by the owner, in Remix)

The mainnet contract is deployed at 0xa0AD40C50001DEFcEe354Ec0316b99F78bD6c64f. To reproduce or redeploy:

  1. Open contracts/src/PocketLM.sol in Remix. Compiler 0.8.24, EVM version paris, optimizer on (200 runs). No constructor arguments.
  2. Deploy with the mainnet wallet on BOT Chain (chain ID 677).
  3. Nothing to set in Vercel: the address above and mainnet are the defaults. Set NEXT_PUBLIC_CONTRACT_MAINNET=<address> only if you deploy a new contract, and NEXT_PUBLIC_MAINNET_RPC=<rpc url> if the default RPC is wrong. Redeploy.
  4. Replace web/public/botchain-logo.svg with the official BOT Chain logo (same file name).

License

The code is released under the MIT License. The PocketLM name and mascot artwork are not covered by it.

Limits, said plainly

  • The score is self-reported. The quiz is graded in the browser, so anyone can stamp any score. A stamp shows that a wallet logged a result at a given time, not that the person mastered the material.
  • The rate limit is a simple per-IP counter in server memory. Fine for a demo; move it to Redis if traffic grows.

Siapa yang menggunakan PocketLM?

Pengguna utama

  1. Mahasiswa dan pelajar (pengguna utama): mengunggah materi kuliah, memahami isinya lewat AI, mengerjakan kuis, dan mencatat progres belajar. Manfaat: belajar lebih cepat dan punya riwayat aktivitas belajar.
  2. Dosen atau mentor (pihak yang memverifikasi): bisa memeriksa study record mahasiswa lewat blockchain explorer, misalnya untuk melihat bahwa mahasiswa mencatat hasil kuis tertentu pada waktu tertentu. Manfaat: memeriksa catatan tanpa bergantung sepenuhnya pada database PocketLM.
  3. Komunitas belajar atau penyelenggara program (pihak yang menerima bukti): program belajar mandiri, komunitas bootcamp, atau penyelenggara challenge belajar yang ingin peserta mencatat progresnya secara transparan. Manfaat: catatan aktivitas belajar yang bisa diperiksa secara independen.

Target pengguna tambahan

  1. Profesional yang melakukan upskilling: karyawan yang mempelajari skill baru (AI, programming, cybersecurity, data analytics). Manfaat: mencatat aktivitas belajar mandiri sebagai pelengkap portofolio.
  2. Kreator kursus dan instruktur online: memakai PocketLM sebagai alat bantu belajar untuk materi yang mereka buat. Integrasi langsung dengan kursus masih perlu dikembangkan.
  3. Perekrut dan tim HR: bisa memeriksa riwayat belajar kandidat sebagai informasi tambahan tentang inisiatif pengembangan diri. Catatan PocketLM bukan bukti kompetensi atau penguasaan skill.
  4. Bootcamp dan penyelenggara pelatihan: mengarahkan peserta mempelajari materi dan mencatat hasil kuis selama program. Verifikasi hasil yang jujur membutuhkan mekanisme tambahan.
  5. Developer dan komunitas Web3: mempelajari dokumentasi blockchain, smart contract, protokol, atau SDK, lalu membagikan study record ke komunitas atau mentor.
  6. Perusahaan dan tim Learning & Development: membantu karyawan memahami SOP, dokumentasi internal, atau materi pelatihan. Untuk dokumen rahasia dibutuhkan kontrol privasi dan akses tambahan.

Roadmap

Versi Fitur dokumen
MVP (V1) 1 dokumen per notebook, PDF atau paste teks
V1.1 Bisa menambahkan beberapa dokumen ke notebook yang sama
V1.2 AI menjawab pertanyaan berdasarkan seluruh dokumen
V2 Ringkasan gabungan, kuis lintas dokumen, dan sitasi sumber dokumen

Hanya MVP (V1) yang dikerjakan sekarang. Versi lain adalah rencana ke depan.

Catatan penting

  • Hanya hash dokumen dan judul yang masuk on-chain. Teks dokumen tidak pernah disimpan.
  • Skor kuis dinilai di browser dan dicatat oleh pengguna sendiri (self-reported). Study record membuktikan bahwa sebuah wallet mencatat hasil pada waktu tertentu, bukan bahwa orang itu menguasai materinya.

About

AI notebook that turns your notes or a short PDF into a summary, key points, a quiz and answers, then stamps your study proof (title, notes hash, score) on BOT Chain.

Topics

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages