Alwakeelo AI

Pakistan's First Open-Source Legal AI Platform
Search 600,000+ judgments · AI legal chat · Draft petitions & contracts · Case management

Website · LinkedIn · Legal Guides · Contact

What is Alwakeelo AI?

Alwakeelo AI is a full-stack legal workspace built for Pakistani advocates, law chambers, and legal professionals. It combines retrieval-augmented generation (RAG) with Pakistan's largest digital judgment database to deliver AI responses grounded in verified case law — eliminating hallucinations and ensuring accurate legal citations.

"Research case law, draft petitions and contracts, and generate client-ready legal documents in minutes with Alwakeelo AI, fine-tuned for Pakistani legal practice."

Features

Alwakeelo ships with 12 core modules:

💬 AI Legal Chat

Consult with an AI legal advisor trained on Pakistani law. Get strategy and next-step guidance fast.

👑 Al Wakeelo Engine

Use the main legal AI workspace with grounded responses, references, and practical next-step guidance.

🔍 Judgment Search

Find relevant Pakistani case law with quick citation-focused search and contextual summaries.

🔗 Citation Search

Search directly by year, journal, and page to locate precise judgments and linked details quickly.

📚 Statute Lookup

Navigate Pakistani statutes and sections with plain-language legal explanations.

📝 Legal Drafting

Prepare petitions, notices, applications, and legal replies with structured templates, clause-ready sections, and style-consistent drafting support.

🧠 Style-Memory RAG

Train AI on your uploads, drafts, and accepted edits so output follows your legal style and preferred language.

📋 Contract Drafting

Generate client-ready contracts with structured clause sets, risk score breakdown, redline suggestions, and cleaner final drafts for negotiation or execution.

📄 Case Documents

Upload, review, and organize matter-specific documents with faster legal analysis support.

📖 Knowledge Vault

Maintain private user documents and global admin legal resources for retrieval-grounded outputs.

👥 Organization Workspace

Support chamber and team workflows with shared access controls and collaboration-ready structure.

🎙️ Audio Transcription

Convert legal voice notes and recorded audio into text for research, drafting, and case preparation.

Additional Case & Calendar Management Features

📁 Case Management

Organize cases by type (criminal, civil, family, constitutional, tax, corporate, banking, labor, property). Track clients/parties with CNIC and contact details, manage 11 compliance types (hearings, filing deadlines, limitation dates, letter of authority, conflict checks, etc.), and link uploaded documents to specific cases.

📅 Daily Diary

Hearing calendar with priority levels, outcomes, next-date tracking, and automated email digests (daily/weekly via Resend). Timezone-aware scheduling (Asia/Karachi default).

Tech Stack

Frontend

Technology	Version	Purpose
React	18.3	UI framework
TypeScript	5.6	Type safety
Vite	7.3	Build tool & dev server
Tailwind CSS	3.4	Utility-first styling
Radix UI	—	Accessible component primitives
TanStack Query	5.60	Server state management & caching
TipTap	—	Rich text editor (13 extensions)
Wouter	3.3	Lightweight client-side routing
Framer Motion	11.18	Animations
Recharts	2.15	Data visualization
react-pdf	10.4	In-browser PDF viewing
react-window	2.2	Virtualized lists
cmdk	1.1	Command palette (⌘K)
Capacitor	8.3	Android native app wrapper
jsPDF	4.2	Client-side PDF generation

Backend

Technology	Version	Purpose
Express	5.0	HTTP server
PostgreSQL	16+	Primary database + pgvector
Drizzle ORM	0.45	Type-safe queries + Zod validation
OpenAI SDK	6.21	AI provider client
@xenova/transformers	2.17	Local ML embeddings (384-dim multilingual)
Passport	0.7	Authentication (email/password + Google OAuth)
openid-client	6.8	Google OpenID Connect
Resend	6.9	Transactional email
Safepay	0.3	Pakistani payment gateway (PKR)
Cloudflare R2	—	S3-compatible object storage
Multer	2.0	File upload handling
Tesseract OCR	—	Local OCR (English + Urdu)
OCR.space	—	Cloud OCR fallback
mammoth	1.11	DOCX → text extraction
unpdf	1.4	PDF text extraction
docx	9.6	DOCX file generation
ws	8.18	WebSocket support

DevDependencies

Technology	Version	Purpose
drizzle-kit	0.31	Database migrations
esbuild	0.25	Server bundling
Playwright	1.59	E2E testing
jsdom	29.1	Unit testing (DOM)
PostCSS + Autoprefixer	—	CSS processing
@tailwindcss/typography	0.5	Prose styling

AI Architecture

Multi-Provider Fallback Chain

User Query
    ↓
Intent Classification → Complexity Detection → Domain Detection
    ↓
Knowledge Retrieval (RAG)
    ├── Case law index (600K+ judgments)
    ├── Statute index (5,900+ indexed statutes)
    ├── Admin knowledge base
    ├── User documents
    └── Style memory (personalization)
    ↓
AI Generation (with fallback chain)
    ↓
Citation Verification → Strip unverified citations
    ↓
Response to user (with clickable, verified citations)

AI Providers & Models

Mode	Primary Model	Fallback Model	Access	Description
Standard	Gemini 3.0 Flash (via OpenRouter)	DeepSeek Chat	All Tiers	Balanced speed and accuracy for standard legal chats and documentation lookup.
Turbo	Kimi K2.5 (`moonshotai/kimi-k2.5` via OpenRouter)	Gemini 3.0 Flash (`google/gemini-3-flash-preview` via OpenRouter)	Pro+	High-speed processing and fluent answers for rapid-fire Q&A. DeepSeek Pro/R1 serves as final fallback.
Apex	Claude 3.5 Sonnet (via OpenRouter)	GPT-4o / DeepSeek R1	Chamber+	Best-in-class legal analysis, drafting, and contract review with support for file context uploads.

The AI router implements a race-with-deadline pattern — if the primary provider fails to emit the first token within the SLA timeout, it automatically cascades to the next provider in the chain. Streaming and non-streaming fallback paths are managed independently.

RAG Pipeline

Component	Provider / Technology	Details
Intent Classifier	Custom Rule + AI	Query complexity, domain detection, and follow-up reference resolution.
Query Rewriter	AI	Reformulates shorthand inputs into detailed queries.
RAG Embeddings	Voyage AI (Primary)	Voyage Law 2 (`voyage-law-2`, 1024-dimension) embeddings for high-quality legal semantic searches.
Reranking	Voyage Reranker	Voyage Rerank 2 (`rerank-2`) for precise contextual scoring.
Vector Store	pgvector	PostgreSQL vectorized storage.
Chunker	Custom sliding-window	Optimizes context window coverage by dividing legal files into overlapping paragraphs.

Database Schema

Alwakeelo maps 45 tables across PostgreSQL with pgvector extension:

Domain	Tables	Highlights
Auth & Sessions	4	Users, sessions (pg-backed), password reset, email verification
Legal Research	8	Judgments (tsvector FTS), case_law, citation_links, unresolved_citations, statutes, statute_documents, law_journals, courts_ref
AI Chat	3	Threads, messages, query_cache
Documents & Files	4	Documents with auto-classification, R2 file metadata
Knowledge (RAG)	4	Admin knowledge, GitHub-synced, org-level, file storage
Style Memory	4	Settings, samples, chunks (pgvector embeddings), events
Case Management	5	Case files, clients/parties, compliance (11 types), notes, diary entries
Organizations	3	Orgs, members (owner/admin/member roles), email invites
Payments	1	Safepay payment records (PKR)
Analytics	4	Usage tracking, AI output quality log, search history, bookmarks
Lead Generation	3	Visitor sessions, case leads, funnel events
Notifications	1	Per-user email digest preferences (daily/weekly)

Subscription Tiers

Tier	Monthly (PKR)	AI Actions	Modes	OCR Pages	Users
Free	0	12	Standard	100	1
Standard	500	120	Standard	250	1
Pro ⭐	1,000	350	Standard + Turbo	500	1
Chamber	4,500	1,200 (pooled)	All modes + Apex	1,500	3
Enterprise	50,000	30,000	All modes + priority	Custom	Custom

Courts & Law Reports

Courts: Supreme Court of Pakistan · Lahore High Court · Sindh High Court · Peshawar High Court · Islamabad High Court · Balochistan High Court · Federal Shariat Court.
Law Reports: PLD · SCMR · YLR · MLD · CLC · CLD · PCrLJ.

SEO Infrastructure

130,000+ indexable pages with server-side SEO:

Server-side meta injection — unique <title>, <meta description>, canonical, and Open Graph/Twitter tags per route via seo-meta.ts.
Pre-render blocks — visible HTML with judgment text, headnotes, and court metadata for crawlers.
Dynamic sitemap — paginated XML sitemaps (10K URLs/page), ordered by decisionDate DESC.
Schema.org — CourtCase, Legislation, BlogPosting, Organization, LegalService, Dataset, WebSite, SoftwareApplication JSON-LD.
IndexNow — real-time Bing/Yandex/Seznam/Naver notification.
Google Indexing API — direct URL submission.
Proper 404s — HTTP 404 for non-existent judgments (prevents soft 404s).
Legacy redirects — 301s for old URLs (/judgment-search → /judgments).

Security

CSRF protection (origin/referer validation on mutating requests).
Content Security Policy (CSP) with strict directives.
Rate limiting (auth, AI, global API — IP-based).
bcrypt password hashing.
CAPTCHA (Cloudflare Turnstile + Google reCAPTCHA fallback).
Optional single-IP session enforcement.
User ban system with admin audit logging.
File upload scanning.
HTTPS enforcement + HSTS in production.
X-Content-Type-Options, X-Frame-Options, Referrer-Policy, Permissions-Policy, COOP, CORP headers.

Getting Started

Prerequisites

Node.js 20+
PostgreSQL 16+ with pgvector extension
Tesseract OCR + poppler-utils (optional, for local OCR)

Installation

# Clone the repository
git clone https://github.com/ijlalbintariq/alwakeeloneon.git
cd alwakeeloneon

# Install dependencies
npm install

# Set up environment variables
cp .env.example .env
# Edit .env with your DATABASE_URL, API keys, etc.

# Push database schema
npm run db:push

# Start development server
npm run dev

Key Environment Variables

Variable	Required / Optional	Description
`DATABASE_URL`	✅ Required	PostgreSQL connection string (with pgvector).
`SESSION_SECRET`	✅ Required	Express session secret.
`DEEPSEEK_API_KEY`	✅ Required	Primary AI provider (DeepSeek v4).
`MOONSHOT_API_KEY`	Optional	API key for direct Moonshot/Kimi integration (points to `https://api.moonshot.ai/v1`).
`OPENROUTER_API_KEY`	Recommended	Fallback AI + Apex mode (Claude Sonnet 5, Kimi K2.5, Gemini 3 Flash).
`GOOGLE_CLIENT_ID`	Optional	Google OAuth client ID for login.
`RESEND_API_KEY`	Optional	Transactional email provider API key (via Resend).
`RESEND_FROM_EMAIL`	Optional	The sender email address for Resend notifications.
`SAFEPAY_API_KEY`	Optional	Public key for Safepay payment gateway integration (PKR checkout session).
`SAFEPAY_SECRET_KEY`	Optional	Secret key for Safepay payment gateway operations.
`SAFEPAY_ENVIRONMENT`	Optional	Environment for Safepay (`sandbox` or `production`).
`SAFEPAY_WEBHOOK_SECRET`	Optional	Webhook verification secret for Safepay payment notifications.
`R2_ACCOUNT_ID`	Optional	Cloudflare R2 Account ID for S3-compatible storage.
`R2_ACCESS_KEY_ID`	Optional	Cloudflare R2 Access Key ID.
`R2_SECRET_ACCESS_KEY`	Optional	Cloudflare R2 Secret Access Key.
`R2_BUCKET`	Optional	Cloudflare R2 Bucket name.
`R2_ENDPOINT`	Optional	Cloudflare R2 endpoint URL.
`R2_REGION`	Optional	Cloudflare R2 region (e.g. `auto`).
`R2_PUBLIC_BASE_URL`	Optional	Public CDN or base URL mapping to Cloudflare R2.
`GOOGLE_INDEXING_CREDENTIALS`	Optional	Google Indexing API JSON credentials for service account URL submission.
`INDEXNOW_KEY`	Optional	Key for IndexNow real-time Bing/Yandex search indexing notification.
`OCRSPACE_API_KEY`	Optional	Cloud OCR.space API key for PDF/image OCR fallback.
`CAPTCHA_ENFORCED`	Optional	Set to `true` to enable CAPTCHA on auth forms.
`TURNSTILE_SECRET_KEY`	Optional	Cloudflare Turnstile CAPTCHA secret key.
`RECAPTCHA_SECRET_KEY`	Optional	Google reCAPTCHA fallback secret key.

Deployment & Production

Docker

docker build -t alwakeelo-ai .
docker run -p 5000:5000 --env-file .env alwakeelo-ai

The Dockerfile uses node:20-bookworm-slim and installs tesseract-ocr (English + Urdu), poppler-utils, and ca-certificates.

Production Building & Starting

npm run build    # Builds client (Vite) + server (esbuild)
npm start        # Runs node dist/index.cjs

NPM Scripts

Script	Purpose
`npm run dev`	Start development server (tsx)
`npm run build`	Production build
`npm start`	Production server
`npm run check`	TypeScript type checking
`npm run db:push`	Push Drizzle schema to database
`npm test`	Unit tests
`npm run test:e2e`	E2E tests (Playwright)
`npm run reindex:admin-caselaw`	Reindex case law RAG vectors
`npm run reindex:statutes`	Reindex statute RAG vectors

Project Structure

├── client/                  # React frontend (Vite)
│   ├── src/
│   │   ├── pages/           # 41 page components
│   │   ├── components/      # 21 shared components + ui/
│   │   ├── hooks/           # Custom React hooks
│   │   └── lib/             # Utilities
│   ├── public/              # Static assets, PWA manifest
│   └── index.html           # SPA shell with schema.org JSON-LD
├── server/                  # Express 5 backend
│   ├── routes.ts            # API routes
│   ├── storage.ts           # Data access layer
│   ├── pipeline/            # AI intent classification & retrieval
│   ├── rag/                 # RAG service, vector store, embeddings
│   ├── style-memory/        # Style learning system
│   ├── retrieval/           # Clause library, TOC parser
│   ├── services/            # Citation extractor, DOCX generator, Google indexing
│   ├── middleware/          # Rate limiting
│   ├── ai-router.ts         # Multi-provider AI with fallback chains
│   ├── deepseek-ai.ts       # DeepSeek AI client
│   ├── openrouter-ai.ts     # OpenRouter AI client
│   ├── apex-ai.ts           # Apex (Claude Sonnet 5) client
│   ├── static.ts            # Static serving + SSR SEO injection
│   ├── seo-meta.ts          # Per-route meta tag generation
│   ├── sitemap.ts           # Dynamic XML sitemap
│   ├── indexnow.ts          # Real-time search engine notification
│   ├── safepay.ts           # Payment processing (PKR)
│   ├── email.ts             # Transactional email (Resend)
│   ├── ocr.ts               # Local Tesseract OCR
│   ├── cloud-ocr.ts         # Cloud OCR.space
│   └── r2-storage.ts        # Cloudflare R2
├── shared/                  # Shared code (client + server)
│   ├── schema.ts            # Drizzle ORM schema (45 tables)
│   ├── blog-data.ts         # Blog articles
│   └── models/              # Auth models
├── Dockerfile               # Production container
├── render.yaml              # Render.com deployment config
├── drizzle.config.ts        # Drizzle migration config
└── vite.config.ts           # Vite build config

Contributing

We welcome contributions! Please open an issue or submit a pull request.

License

This project is licensed under the MIT License.

Contact

Website: www.alwakeelo.com
Email: support@alwakeelo.com
Phone/WhatsApp: +92 335 834 1897
LinkedIn: Al Wakeelo

Built with ❤️ in Pakistan 🇵🇰

Name		Name	Last commit message	Last commit date
Latest commit History 1,074 Commits
.aidesigner		.aidesigner
.claude		.claude
.codex		.codex
.local/state/replit/agent		.local/state/replit/agent
attached_assets		attached_assets
client		client
docs		docs
migrations		migrations
reports		reports
scratch		scratch
script		script
scripts		scripts
server		server
shared		shared
tests		tests
.buildinfo		.buildinfo
.cachebust		.cachebust
.env.example		.env.example
.gitignore		.gitignore
.npmrc		.npmrc
.replit		.replit
2026_only.jsonl		2026_only.jsonl
CHAT_PERFORMANCE_ANALYSIS.md		CHAT_PERFORMANCE_ANALYSIS.md
CITATION_FIX_GUIDE.md		CITATION_FIX_GUIDE.md
DEVELOPER_GUIDE.md		DEVELOPER_GUIDE.md
Dockerfile		Dockerfile
JUDGMENT_VAULT_PLAN.md		JUDGMENT_VAULT_PLAN.md
LIVE_PRODUCTION_TEST_REPORT.md		LIVE_PRODUCTION_TEST_REPORT.md
MANUAL_TEST_CHECKLIST.md		MANUAL_TEST_CHECKLIST.md
ORIGINAL_REQUEST.md		ORIGINAL_REQUEST.md
PROJECT.md		PROJECT.md
RAG_FORENSIC_AUDIT_REPORT.md		RAG_FORENSIC_AUDIT_REPORT.md
README.md		README.md
REWRITE_DOCUMENTATION.md		REWRITE_DOCUMENTATION.md
SECURITY.md		SECURITY.md
TEST_VERIFICATION.md		TEST_VERIFICATION.md
ai-model-test.js		ai-model-test.js
ai-model-test.mjs		ai-model-test.mjs
ai-model-test.ts		ai-model-test.ts
all_modes_diverse_results.json		all_modes_diverse_results.json
build_output.log		build_output.log
chatgpt-app-submission.json		chatgpt-app-submission.json
components.json		components.json
contract-draft-test.mjs		contract-draft-test.mjs
detailed_forensic_audit_report.md		detailed_forensic_audit_report.md
drizzle.config.ts		drizzle.config.ts
judgment_ingestion_walkthrough.md		judgment_ingestion_walkthrough.md
live-production-test.mjs		live-production-test.mjs
live-test.mjs		live-test.mjs
package-lock.json		package-lock.json
package.json		package.json
parsed_sections.json		parsed_sections.json
pipeline_test.ts		pipeline_test.ts
postcss.config.js		postcss.config.js
progress.md		progress.md
render.yaml		render.yaml
replit.md		replit.md
scratch_existing_citations.json		scratch_existing_citations.json
search_results.txt		search_results.txt
statute_seeding_walkthrough.md		statute_seeding_walkthrough.md
statute_test_report.md		statute_test_report.md
tailwind.config.ts		tailwind.config.ts
test-safepay.mjs		test-safepay.mjs
tsconfig.json		tsconfig.json
vite.config.ts		vite.config.ts
walkthrough.md		walkthrough.md

Folders and files

Latest commit

History

Repository files navigation

Alwakeelo AI

What is Alwakeelo AI?

Features

💬 AI Legal Chat

👑 Al Wakeelo Engine

🔍 Judgment Search

🔗 Citation Search

📚 Statute Lookup

📝 Legal Drafting

🧠 Style-Memory RAG

📋 Contract Drafting

📄 Case Documents

📖 Knowledge Vault

👥 Organization Workspace

🎙️ Audio Transcription

Additional Case & Calendar Management Features

📁 Case Management

📅 Daily Diary

Tech Stack

Frontend

Backend

DevDependencies

AI Architecture

Multi-Provider Fallback Chain

AI Providers & Models

RAG Pipeline

Database Schema

Subscription Tiers

Courts & Law Reports

SEO Infrastructure

Security

Getting Started

Prerequisites

Installation

Key Environment Variables

Deployment & Production

Docker

Production Building & Starting

NPM Scripts

Project Structure

Contributing

License

Contact

About

Topics

Resources

Security policy

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Contributors

Uh oh!

Languages

Packages