Every feature in LegalMind was designed around real student and advocate workflows — not adapted from a generic AI tool.
Designed around real student and advocate workflows — not a generic AI wrapper with a legal skin.
Browse thousands of digitised statutes, case law, and commentaries — parsed to structured markdown with full-text search and section-level navigation.
Curated PYQ bank with model answers, filterable by subject, year, and examination — perfect for bar exam and competitive preparation.
Long-press any passage to highlight it in one of five colours. All annotations sync across devices and surface in a consolidated bookmarks panel.
Generate A4-formatted printouts of any document with your institution watermark, configurable page range, and share to PDF or printer.
Hybrid semantic + keyword retrieval across your entire library. Every answer is anchored to exact citation paragraphs — no hallucinations.
A deterministic, auditable pipeline — every step is locally inspectable and reproducible from source.
Upload a PDF, DOCX, or scanned statute. The pipeline extracts text, detects structure — chapters, sections, definitions, schedules — and produces canonical, clean markdown.
Supports PDF, DOCX, scanned images, and court judgement formats.
Each section is chunked at semantic boundaries, embedded into pgvector, and indexed with confidence scoring. Low-confidence pages go to FastVLM for visual understanding.
Every parse is versioned. Nothing overwrites the original.
Searches combine semantic vector similarity and keyword BM25. Results are ranked and anchored to exact citation paragraphs — not just documents.
Citation anchors point to the specific paragraph, not just the document.
The local LLM generates answers strictly from the retrieved passages. No internet access. No invented references. Every claim points back to a real paragraph you can verify.
Runs entirely on your machine via Ollama. Your documents never leave your device.