Bright Data Web Data UNLOCKED Hackathon 2026 — Submission · Deadline: May 30, 2026
A 5-agent AI pipeline that converts a plain-language legal situation into a complete, jurisdiction-specific action plan — statutes, case precedents, a drafted complaint letter, the exact filing portal, and legal aid contacts — in under 30 seconds. 4 Bright Data API calls per query. Powered entirely by Bright Data's live web infrastructure.
- ✅ Track 3: Security & Compliance (primary)
- ✅ Track 1: GTM Intelligence (secondary)
- ✅ Partner: AI/ML API Challenge
- ✅ Partner: Speechmatics Challenge
- ✅ Bright Data Requirement: SERP API + Browser API integrated
| Judging Criteria | How LexScout Delivers |
|---|---|
| Bright Data Integration | 4 Bright Data API calls per query — SERP API (law search) + Browser API (case scraping + portal extraction + legal aid navigation) — both live |
| Technical Complexity | 5-agent LangGraph pipeline, async FastAPI, real-time web scraping, Groq + AI/ML API dual-LLM architecture |
| Real-World Impact | 1.4B addressable users, $0 cost, 3 jurisdictions live on day one |
| Innovation | First pipeline to combine live statute search + precedent scraping + auto-complaint generation in a single API call |
| Demo Quality | 3 live scenarios across India, USA, EU with real queries and real Bright Data responses |
85% of legal violations go unreported. $300+ per lawyer consultation. 40M+ pending cases — citizens don't know where to start.
A domestic worker photographed without consent doesn't know Section 66E exists. A gig worker whose wages were illegally deducted doesn't know which Labour Commissioner to file with. A consumer defrauded online doesn't know they can file on consumerhelpline.gov.in for free.
The gap: Rich legal data exists publicly. A tool that turns it into action doesn't.
Describe your situation in plain language. LexScout returns a complete legal action plan in ~20 seconds:
| Output | Example |
|---|---|
| 📜 Relevant Laws | 2–4 statutes — Payment of Wages Act 1936 §7; Industrial Disputes Act 1947 §25F |
| ⚖️ Case Precedents | 3–5 cases — Workmen v. Reptakos Brett & Co. (1992) — employer liable for unlawful deductions |
| 📝 Complaint Letter | Full formal letter with legal citations, ready to file |
| 🌐 Filing Portal | https://clc.gov.in/ — live-scraped, verified link |
| 🤝 Legal Aid | NALSA + local orgs with phone, email, website |
| 🗺️ Action Plan | 5 concrete next steps ordered by urgency |
Free. Instant. No lawyer needed.
Homepage — Describe your situation
Query input with hint chips, jurisdiction selector, mic button, and stats bar
Results — Laws & Case Precedents
Applicable laws with timeline, relevant case precedents with outcomes
Action Plan — Complaint Letter, Filing Portal & Legal Aid
Auto-drafted complaint letter citing exact laws, official government filing portal, legal aid contacts
Tech Stack — Groq + AI/ML API + Bright Data + LangGraph
Full technology stack with live request flow: Voice/Text → Speechmatics → 5 Agents → Bright Data SERP/Browser → Groq llama-3.3-70b + Mistral-7B → Legal Action Plan
Impact & Prize Tracks
1.4B potential users · <30s response · $0 cost · 3+ jurisdictions · $2,400+ prize track eligibility
Static legal databases go stale within months. Laws get amended. Complaint portals change URLs. Case databases add thousands of new judgments daily. LexScout cannot exist without live web data — Bright Data is the infrastructure that makes this possible.
4 Bright Data API calls per query, each mapped to a dedicated agent in the pipeline:
| Agent | Law Finder (Agent 2) |
| Targets | indiacode.nic.in · law.cornell.edu · eur-lex.europa.eu |
| What it does | Searches government legal databases for statutes matching the user's legal category and jurisdiction |
| Why Bright Data | Legal government portals block generic scrapers with rate limits and geo-restrictions. SERP API surfaces authoritative results reliably |
| Output | Raw results → Groq extracts {law_name, article, created, last_amended, covers} — 2–4 laws per query |
| Agent | Precedent Hunter (Agent 3) |
| Targets | indiankanoon.org · courtlistener.com · curia.europa.eu |
| What it does | Fetches full rendered page content from case law databases behind Cloudflare, CAPTCHAs, and browser fingerprinting |
| Why Bright Data | Indian Kanoon and CourtListener implement heavy bot-detection. Web Unlocker handles JS rendering, fingerprint rotation, and residential IP routing |
| Output | Raw HTML → Groq parses {case_name, outcome, year, relevance, citation} — 3–5 precedents per query |
| Agent | Action Builder (Agent 4) — portal step |
| Targets | consumerhelpline.gov.in · ftc.gov · cybercrime.gov.in |
| What it does | Navigates government complaint portals that render forms dynamically via JS and extracts the exact filing URL |
| Why Bright Data | Government portals load complaint URLs dynamically — a static HTTP request returns a blank page. Scraping Browser renders them fully |
| Output | portal_url — the verified, live complaint filing link |
| Agent | Action Builder (Agent 4) — legal aid step |
| Targets | nalsa.gov.in/lsas · lawhelp.org · e-justice.europa.eu |
| What it does | Multi-step browser navigation: search → filter by category → paginate → extract contact info |
| Why Bright Data | Legal aid directories require real browser navigation. Static scrapers fail completely |
| Output | legal_aid_contacts[] — list of {organization, phone, email, website, free_service} |
LexScout supports voice-to-text query input via the Speechmatics Real-time API (wss://eu2.rt.speechmatics.com/v2).
- Click the 🎤 mic button next to the search box
- MediaRecorder streams audio as
audio/webm;codecs=opusover WebSocket - Speechmatics transcribes in real time using the
enhancedoperating point - Transcript is pasted directly into the query textarea as you speak
- Red pulse animation shows active recording state; click again to stop
This makes LexScout accessible to users who cannot type — particularly important for the rural and low-literacy demographics who face the most legal vulnerability.
Groq powers all reasoning in Agents 1–4 of the pipeline:
| Agent | Task |
|---|---|
| Classifier | Identify legal category + jurisdiction from plain-language input |
| Law Finder | Parse Bright Data SERP results → structured law objects |
| Precedent Hunter | Parse SERP snippets → structured case objects |
| Action Builder | Draft jurisdiction-specific complaint letter with exact citations |
The executive summary (Agent 5 — Synthesizer) is generated via the AI/ML API using mistralai/Mistral-7B-Instruct-v0.2:
- Endpoint:
https://api.aimlapi.com/v1/chat/completions - Produces: 3-sentence executive summary, urgency score, 5-step action plan
- Fallback: Groq llama-3.3-70b if AI/ML API is unavailable
User (browser) ──🎤 Speechmatics RT API──► textarea
│
▼ POST /query {"query": "...", "country": "India"}
│ ~20s end-to-end
▼
┌──────────────────────────────────────────────────────────────┐
│ LexScout StateGraph (LangGraph) │
│ │
│ ┌─────────────────┐ │
│ │ Agent 1 │ │
│ │ Classifier │ Groq llama-3.3-70b │
│ │ │ → category, jurisdictions │
│ └────────┬────────┘ │
│ ▼ │
│ ┌─────────────────┐ │
│ │ Agent 2 │ bright_data_search() │
│ │ Law Finder │ ← Bright Data SERP API │
│ │ │ Targets: indiacode.nic.in │
│ │ + Groq │ law.cornell.edu │
│ │ → laws[] │ eur-lex.europa.eu │
│ └────────┬────────┘ │
│ ▼ │
│ ┌─────────────────┐ │
│ │ Agent 3 │ bright_data_access() │
│ │ Precedent Hunter│ ← Bright Data Web Unlocker │
│ │ │ Targets: indiankanoon.org │
│ │ + Groq │ courtlistener.com │
│ │ → precedents[] │ curia.europa.eu │
│ └────────┬────────┘ │
│ ▼ │
│ ┌──────────────────────────────────────────────────┐ │
│ │ Agent 4 │ │
│ │ Action Builder │ │
│ │ │ │
│ │ bright_data_extract() ← Scraping Browser │ │
│ │ → portal_url (live complaint filing link) │ │
│ │ │ │
│ │ bright_data_interact() ← Browser Automation │ │
│ │ → legal_aid_contacts[] (structured orgs) │ │
│ │ │ │
│ │ Groq llama-3.3-70b → complaint_draft │ │
│ └────────┬─────────────────────────────────────────┘ │
│ ▼ │
│ ┌─────────────────┐ │
│ │ Agent 5 │ AI/ML API Mistral-7B │
│ │ Synthesizer │ → executive_summary │
│ │ │ → next_steps[] │
│ │ │ → urgency │
│ └────────┬────────┘ │
└───────────┼──────────────────────────────────────────────────┘
▼
┌──────────────────────────────────────────────────────────────┐
│ Final JSON Response │
│ laws[] · precedents[] · complaint_draft · portal_url │
│ legal_aid[] · executive_summary · next_steps[] · urgency │
└──────────────────────────────────────────────────────────────┘
▼
HTML Frontend — dark theme, animated, mobile-responsive
5 collapsible sections · staggered fade-in · voice input
| Layer | Technology | Purpose |
|---|---|---|
| Data | Bright Data SERP API | Live legal statute search |
| Data | Bright Data Web Unlocker | Case law database access |
| Data | Bright Data Scraping Browser | Portal URL extraction |
| Data | Bright Data Browser Automation | Legal aid directory navigation |
| Voice | Speechmatics Real-time API | WebSocket speech-to-text |
| LLM | Groq llama-3.3-70b | Agents 1–4: classification, parsing, drafting |
| LLM | AI/ML API — Mistral-7B | Agent 5: executive summary synthesis |
| Orchestration | LangGraph StateGraph | 5-agent pipeline with shared state |
| Backend | FastAPI + Uvicorn | REST API, CORS, request validation |
| Validation | Pydantic v2 | Request/response schema |
| Frontend | Pure HTML5 + CSS3 + Vanilla JS | Zero-dependency UI, dark theme |
| Fonts | Google Fonts — Syne, DM Mono | Typography |
| Deployment | Railway.app | Live public URL for submission |
| Language | Python 3.11 | Windows 11 + PowerShell |
| Dimension | Values |
|---|---|
| Jurisdictions | 🇮🇳 India · 🇺🇸 USA · 🇪🇺 EU |
| Legal Categories | Privacy · Employment · Consumer Rights · IP · Criminal |
| Law Databases | indiacode.nic.in · law.cornell.edu · eur-lex.europa.eu |
| Case Databases | indiankanoon.org · courtlistener.com · curia.europa.eu |
| Filing Portals | cybercrime.gov.in · consumerhelpline.gov.in · ftc.gov · clc.gov.in + more |
| Legal Aid | nalsa.gov.in · lawhelp.org · e-justice.europa.eu |
lexscout/
├── backend/
│ ├── agents.py # LangGraph StateGraph — all 5 agent nodes
│ ├── main.py # FastAPI app — POST /query endpoint
│ ├── tool_functions.py # 4 Bright Data async tool functions
│ ├── langgraph_agent.py # LangChain @tool wrappers + graph definition
│ ├── mcp_server.py # MCP server exposing 4 BD tools over stdio
│ ├── requirements.txt # Pinned Python dependencies
│ ├── Procfile # Railway/Render deployment
│ ├── railway.toml # Railway IaC config
│ └── render.yaml # Render.com IaC config
├── frontend/
│ ├── index.html # Main UI — voice input, 5 result sections
│ ├── pitch.html # 5-slide pitch deck
│ └── screenshots/ # UI screenshots for README
├── .env.example # Environment variable template
├── .gitignore
├── LICENSE
└── README.md
git clone https://github.com/ashish-doing/lexscout
cd lexscout && cp .env.example backend/.env
# Add your API keys to backend/.env
cd backend && pip install -r requirements.txt
uvicorn main:app --reload
# Open frontend/index.html in browser- Python 3.11+
- Groq API key (free tier works)
- AI/ML API API key
- Speechmatics API key
- Bright Data account with SERP and Browser zones
GROQ_API_KEY=your_groq_api_key_here
AIML_API_KEY=your_aiml_api_key_here
SPEECHMATICS_API_KEY=your_speechmatics_api_key_here
BRIGHT_DATA_API_KEY=your_bright_data_api_key_here
BRIGHT_DATA_SERP_ZONE=serp_api1
BRIGHT_DATA_BROWSER_ZONE=browser_zone1
ENV=development
PORT=8000curl -X POST http://localhost:8000/query `
-H "Content-Type: application/json" `
-d '{"query": "Someone took my photo without permission in a restaurant. I am in Delhi.", "country": "India"}'Request:
{
"query": "My employer deducted salary without notice. I work in Mumbai.",
"country": "India"
}Response (200):
{
"laws": [{"law_name": "Payment of Wages Act, 1936", "article": "Section 7", "...": "..."}],
"precedents": [{"case_name": "Workmen v. Reptakos Brett & Co. (1992)", "...": "..."}],
"action_plan": {
"complaint_draft": "To The Regional Labour Commissioner...",
"portal_url": "https://clc.gov.in/",
"legal_aid_contacts": [{"organization": "NALSA", "phone": "15100", "...": "..."}],
"next_steps": ["Step 1: Document all deductions...", "..."]
},
"meta": {
"bright_data_tools_used": [
"bright_data_search → SERP API",
"bright_data_access → Web Unlocker",
"bright_data_extract → Scraping Browser",
"bright_data_interact → Browser Automation"
],
"elapsed_seconds": 18.4
}
}| Method | Path | Description |
|---|---|---|
GET |
/ |
Root health check |
GET |
/health |
Health + API key status |
GET |
/supported |
Jurisdictions, categories, data sources |
GET |
/docs |
Swagger interactive docs |
- Push to GitHub
- railway.app → New Project → Deploy from GitHub →
ashish-doing/lexscout - Railway reads
backend/railway.tomlautomatically - Add env vars:
GROQ_API_KEY+AIML_API_KEY+SPEECHMATICS_API_KEY+BRIGHT_DATA_API_KEY+ zone names - Live URL: https://lexscout-production.up.railway.app
render.com → New Web Service → connect repo → Render detects backend/render.yaml → add env vars → deploy.
▶ Watch the 4-minute demo on YouTube
| Metric | Value |
|---|---|
| Addressable users | 500M+ people who face legal issues annually without counsel |
| Cost reduction | From ₹3,000–₹15,000/hr → ₹0 |
| Time to action | From weeks (lawyer consultation) → ~20 seconds |
| Scale path | 50+ jurisdictions via Bright Data's global proxy network |
| B2B path | API for legal aid NGOs, court assistance programs, HR platforms |
| Accessibility | Voice input via Speechmatics for low-literacy and rural users |
Ashish Kumar — B.Tech ECE, IIIT Guwahati (Batch 2024)
MIT — see LICENSE for details.
Built for the Bright Data Web Data UNLOCKED Hackathon — May 2026
Powered by Bright Data · Speechmatics · Groq · AI/ML API · LangGraph
⚖️ Know your rights. Take action.