Automated preliminary Indian property title screening engine. Features multilingual OCR extraction across Indian scripts, deed chain verification, timeline gap detection, tax mutation validation, and advocate briefing pack generation.
Automated legal document ingestion and cross-reference verification.
Tesseract + EasyOCR + Gemini Vision ensemble extracts text, stamp paper seals, and registration stamps across English, Tamil, Telugu, Kannada, Hindi, and Marathi.
Cross-document correlation of Survey Numbers, Sub-division, Patta, Village, Taluk, Boundaries (Four-boundaries verification), and party name variations.
Constructs the full chain of ownership across 30+ years. Detects unrecorded partitions, unregistered power of attorneys (PoA), and registration date lags.
Generates a structured, concise legal briefing document highlighting missing EC/Patta certifications, boundary mismatches, and specific inquiry questions for title certification.
Engineered in Python with LangGraph and Streamlit.
| Layer | Component | Responsibility |
|---|---|---|
| Workflow Engine | LangGraph 1.2, Python 3.11 | Stateful DAG execution for legal validation and cross-doc checking |
| Vision & OCR | Tesseract, EasyOCR, Poppler, Gemini Vision | High-accuracy multi-lingual document extraction and quality scoring |
| User Interface | Streamlit 1.40 | Interactive intake, document viewer, and timeline explorer |
| Portal Verification | SSRF-hardened HTTP client | Reachability testing against state land-records portals |