The problem Argus solves

Project Argus solves flawed legal petitions with dead citations and weak evidence.

Lawyers upload case files: petitions, contracts, exhibits and the 12-stage AI pipeline instantly:

✅ Detects overruled precedents like NN Global II
✅ Weaponizes overlooked CFO emails proving fraud knowledge
✅ Reveals S.447→PMLA asset freeze connections
✅ Predicts judge arbitration bias and opponent moves
✅ Delivers 7-step strategy boosting case score 3.8→7.5/10

Replaces weeks of junior associate research with:

• 10x faster analysis vs manual Westlaw searches
• 100% safer verified citations, zero hallucinations
• Live interactive tabs tracking AI thinking in real-time
• Court-ready petitions that actually win cases

Challenges we ran into

Challenges: Citation Verification & Web Scraping Reliability

Bug: Indian Kanoon/MCA sites frequently block scrapers (403 errors)
Hurdle: 30% false negatives - missed critical precedents/defaulter data

Solutions:
✅ Brightdata proxies + rotating user agents (95% uptime)
✅ Dual verification - Kanoon + SCC Online cross-checks
✅ 24h MongoDB cache prevents re-scraping same URLs
✅ Fallback APIs - Surepass MCA when scraping fails
✅ Confidence scoring - unverified citations flagged <70%

Result: Zero hallucinations, 100% citation traceability maintained